Malay Mail - Tech-gadgets

China shifts AI data centres west as computing demand fuels massive expansion

Thu, 27 Aug 2026 07:00:00 +0800

BEIJING, Aug 27 — In China’s hilly Guizhou province, a cluster of European-style buildings complete with clock tower and multi-arched bridge emits a low, permanent hum—a clue to its unexpected identity as tech giant Huawei’s largest data centre.

The town-like complex is unique in design, but not in its placement. Dozens of data centres have been constructed in the southwestern province in recent years.

They are part of a government strategy that has seen much of the digital infrastructure for breakneck AI development across China’s eastern seaboard established in the country’s less populated, rural west.

The build-out is ongoing. China is on course to nearly double its data centre capacity within the next five years, as a tech race with the United States intensifies and countries everywhere try to swerve a computing crunch that could stymie AI expansion.

The projects have transformed some places—such as Guian New Area, where the Huawei data centre and others are located.

“It used to be a barren mountain, but since the area has developed, transportation has become more convenient, trade has picked up, business opportunities have emerged,” local shopkeeper Shu Peihua said.

When AFP visited last month, street vendors were catering to data centre workers on their lunch break near a roadside banner calling for “promoting high-quality development”, a government tech policy catchphrase.

In Europe and the United States, data centre construction has often been met with resistance from local communities citing concerns around energy prices, as well as environmental issues such as water usage and noise.

Dissent in China is tightly controlled, however, especially around central government policy.

Still, Guian residents who spoke to AFP were positive about the changes to their area.

A vendor selling savoury pancakes, who did not give his name, said two universities had been established since the facilities were built.

Another local, Li Xixiu, said the main road had been developed because of the centres, which in general had “really boosted the economy of this entire area”.

“All the villagers in our neighbourhood now find jobs nearby, unlike before, when they had to go to other places,” Li said.

East to west

In 2021, the government gave its strategy a name: “Eastern Data Western Computing”.

The logic of construction in places like Guizhou, with its cooler climates, abundant renewable energy and empty space, is multifaceted.

AI processing is extremely energy intensive, and China has demanded that data centres draw 80 per cent of their power from renewable sources by the end of this decade.

That makes regions like Guizhou and Inner Mongolia, which have massively expanded renewable capacity like solar and wind, ideal for operators looking to meet those commitments, said Simeng Deng, a senior analyst at Rystad Energy.

Energy-guzzling server farms could also help solve renewables’ issue of curtailment, or the over-production of energy.

The data centres “can be an engine to digest or absorb the surplus of renewable power generation”, said Deng.

Developing previously neglected parts of the countryside is another potential benefit.

Despite the enthusiasm of the locals AFP met though, some researchers question the overall economic gains to host regions.

Data centres do not necessarily create a “huge spillover effect” on local jobs, said Andrew Stokols from Singapore Management University.

“The hope is that you can use data centres as an engine for broader digital development... bringing in industries that will come in around data centres,” he said.

However, both in China and elsewhere, “immediate benefits have been elusive”.

‘Limited’ economic boost

Guizhou has averaged an annual GDP growth rate of 7.4 per cent over the past decade, but its wage growth over the same period was second to last nationwide, according to the Taiwan-based Research Institute for Democracy, Society and Emerging Technology (DSET).

“This contrast suggests that while data centres drive heavy investment in land and equipment, their impact on boosting local per capita income remains limited,” DSET researchers Cartus Bo-Xiang You and Angela Glowacki told AFP by email.

Massive investment in infrastructure, coupled with policy incentives to attract data centre builders, have earned Guizhou one of the highest debt burdens in the country.

In 2024, Guizhou ranked 30th among Chinese provinces for its debt-to-revenue ratio, according to a DSET report, and 27th for its debt-to-GDP ratio.

However, the demand for data processing capacity as countries scramble for technological supremacy means the building will continue.

Even before ChatGPT heralded the beginning of the AI boom, Chinese policy documents were “talking about the idea that computing power is correlated with national power”, said Stokols.

“It’s essential for national competitiveness in the age of digital technology... the AI competition has accelerated that.” — AFP

Mac Mini 2026 Malaysia: M6 and M5 Pro, priced from RM3,799

Alexander Wong — Wed, 26 Aug 2026 11:13:43 +0800

KUALA LUMPUR, Aug 26 — Ahead of the new iPhone 18 Pro announcement, Apple has announced its updated Mac Mini for 2026. While it retains the same compact 5-inch by 5-inch desktop form factor, the new Mac Mini gets significant performance upgrade with the new M6 and M5 Pro chips.

However, the starting price has gone up significantly due to the current global RAM shortage. The new Mac Mini with M6 starts from RM3,799 in Malaysia, which is RM1,300 more than the M4 Mac Mini’s RM2,499 launch price in 2024.

Besides the newer processors, Apple has upgraded wireless connectivity with WiFi 7 and Bluetooth 6, while wired LAN is now faster with 2.5Gb Ethernet.

Mac Mini 2026 Malaysia pricing and availability

The new Mac Mini will be available for pre-order in Malaysia starting Thursday, 27 August 2026 at 9am MYT through the Apple online store and Apple Store app. It will be available from 22 September through Apple Store locations and Apple Authorised Resellers.

In Malaysia, the Mac Mini with M6 starts from RM3,799, while the more powerful Mac Mini with M5 Pro starts from RM7,499.

The base M6 configuration comes with a 12-core CPU, 12-core GPU, 16GB of unified memory and 256GB SSD storage.

Meanwhile, the base M5 Pro configuration gets a 15-core CPU, 16-core GPU, 24GB of unified memory and 512GB SSD storage.

If money is no object, the Mac Mini with M5 Pro can be configured with an 18-core CPU, 20-core GPU, 64GB of unified memory, 8TB SSD and 10Gb Ethernet. This fully-specced configuration costs RM30,449.

Education pricing is also available, starting from RM3,399 for the M6 model and RM7,099 for the M5 Pro model.

What’s new with Mac Mini 2026?

The biggest upgrade for the new Mac Mini is under the hood. The M6-powered model now gets a 12-core CPU and 12-core GPU, which is a significant upgrade from the base 10-core CPU and 10-core GPU option on the previous M4 model.

According to Apple, the M6 delivers up to 40 per cent faster CPU performance and up to 2x faster graphics compared to the M4. Its GPU also gets Neural Accelerators in each core, while the chip features a Dual 16-core Neural Engine. Apple claims this provides up to 4x faster AI performance compared to the previous M4-powered Mac Mini.

Apple says the M6 can deliver up to 4.8x faster LLM prompt processing in LM Studio and up to 1.5x faster spreadsheet calculations in Microsoft Excel compared to the M4. For gaming, Apple claims up to 2x faster ray-tracing performance in Cyberpunk 2077: Ultimate Edition.

The M6 Mac Mini still starts with 16GB of unified memory and can be configured with up to 32GB. Maximum SSD storage also remains at 2TB. However, memory bandwidth has increased from 120GB/s on the M4 to as much as 170GB/s on the M6.

For heavier workloads, the M5 Pro Mac Mini offers up to an 18-core CPU and 20-core GPU. In comparison, the previous M4 Pro topped out with a 14-core CPU and 20-core GPU.

The unified memory can be upgraded up to 64GB, while maximum SSD storage is still 8TB. Memory bandwidth has increased from 273GB/s on the M4 Pro to 307GB/s on the M5 Pro.

Apple claims the M5 Pro delivers up to 4x faster LLM prompt processing in LM Studio, up to 1.4x faster ray-traced rendering in Blender and up to 1.5x faster image processing in Affinity compared to the M4 Pro.

There are also some useful connectivity upgrades across the range. While the previous Mac Mini came with Gigabit Ethernet as standard with 10Gb Ethernet available as an option, the 2026 models now get faster 2.5Gb Ethernet as standard. There’s also the option to upgrade to 10Gb Ethernet if required.

Wireless connectivity has also been upgraded from WiFi 6E to WiFi 7 and Bluetooth 5.3 to Bluetooth 6.

At the front, you still get two USB-C ports supporting USB 3 speeds of up to 10Gbps and a 3.5mm headphone jack. Around the back, there are three Thunderbolt ports, HDMI and Ethernet. The M6 version uses Thunderbolt 4 with speeds up to 40Gbps, while the M5 Pro gets Thunderbolt 5 with speeds up to 120Gbps.

The M6 Mac Mini supports up to three external displays, including two displays at up to 6K 60Hz plus a third display at up to 5K 60Hz. The M5 Pro model can drive up to three 6K 60Hz displays. — SoyaCincau

China’s Bilibili video app is going global — could it be YouTube’s true rival?

Erna Mahyuni — Fri, 21 Aug 2026 14:50:31 +0800

KUALA LUMPUR, Aug 21 — If you follow Chinese social media you might already have heard of BiliBili, China’s answer to YouTube.

On August 19, the service confirmed it was going international, with an Android app for global users already being available while an iOS app is coming soon.

A Discord account has been made to help educate users on how to post their videos on the website and marketing itself by saying that the site aims to reach “Gen Z Coded, affluent and well-educated” users.

What is BiliBili?

It’s a video sharing website. Think YouTube but China-based and until now, mostly skewed towards users in Mainland China.

In recent years it started offering paid subscriptions with premium content including a decent library of Japanese anime as well as China-made animated series known as donghua.

Isn’t it already available internationally?

The previous international version of the app on Android was discontinued but has now been relaunched.

Bilibili has also now removed the need for ID verification, which was a previous requirement for non-China users to upload videos to the platform.

Is it really that popular?

According to Nanjing Marketing Group, in 2025 BiliBili had approximately 340 million monthly active users (MAU) and 34 million daily active users (DAU).

For video content creators who have found YouTube far too crowded or would like an international audience not gatekept by Google’s rules, the numbers are tempting.

Getting into it now while it’s still in the early days of expansion worldwide does look good on paper but with certain caveats.

China site, China rules

In BiliBili’s community rules, it’s stated: “We encourage the creation of new and diverse content. However, we must also take all necessary measures to stop and prevent violations and/or infringement of any laws or regulations, ethics, the requirements of local regulatory agencies, and the legal rights of others, as well as the distribution of any content that may affect the safety and health of the BiliBili community.”

You could interpret this to mean that BiliBili is likely to adhere to China’s general content standards, seeing that it is headquartered in China.

Should I get into BiliBili?

If your content’s growth is stagnating or you think you want to expand your reach globally, it looks good on paper.

The reality is that YouTube’s existing platform, tools and monetisation methods are a lot more straightforward if you’re not based in China.

BiliBili is also likely to have a lot more content moderation with content that YouTube might accept being blocked or flagged instead on the service.

Prospective users should probably get on the site’s official Discord Creator Hub where you can ask questions or at least get a feel for whether the service is the right fit for you.

You could also just head to either the original Chinese website at www.bilibili.com or the more international user-friendly www.bilibili.tv.

A very Meta medieval move: Mark Zuckerberg buys three-storey Gothic Irish castle

Fri, 21 Aug 2026 09:25:48 +0800

DUBLIN, Aug 21 — Tech billionaire Mark Zuckerberg and his wife Priscilla Chan have bought an Irish castle and are “excited” about the historic home, a spokesman for the businessman said in a statement yesterday.

Meta’s founder and CEO bought the three-storey Gothic-style Strancally Castle from the Alen-Buckley family in recent weeks.

“Mark and his family are excited to continue caring for this historic home and look forward to spending time in Ireland, where Meta maintains its international headquarters,” the spokesman said.

Built in 1827 for an English politician, the property is a three-storey, cut-stone Gothic Revival house in County Waterford, about 150 kilometres (93 miles) southwest of Dublin.

The price paid by Zuckerberg in the off-market deal is not known.

The property has been extensively renovated over the past 20 years by its previous owners.

Michael Alen-Buckley is a co-founder of London-based private investment firm RAB Capital.

Meta employs about 1,500 people in Ireland, with Dublin home to its international headquarters.

Its Irish operations also include a virtual and augmented reality business in Cork and a data centre outside Dublin.

Zuckerberg and his family will continue to live in the United States and use the Irish property when conducting Meta business in Ireland. — AFP

‘GTA VI’ isn’t at Gamescom but it’s still dominating the show

Fri, 21 Aug 2026 09:01:21 +0800

PARIS, Aug 21 — The videogames industry meets on Tuesday in Cologne at the Gamescom trade show as concerns mount for its future amid job cuts, cut-throat competition and the end of discs.

Though the latest instalment in the smash-hit Grand Theft Auto series will not be on show at Gamescom, developers are under pressure to make their games stand out from the crowd as developers bring releases forward to try and avoid the GTA behemoth.

After a launch event showcasing the main upcoming releases — with the latest instalment in the Final Fantasy saga as the centrepiece — the gaming gathering opens its doors to industry on Wednesday and to the public on Thursday, running through Sunday.

About 360,000 people attended last year’s jamboree.

In the shadow of GTA VI

The sixth instalment in the hit Grand Theft Auto series — likely to smash records when it comes out on November 19 — is giving rival publishers nightmares.

Other studios have pulled releases forward to try and avoid clashing with Rockstar Games’ GTA release which has so far enjoyed an “unprecedented and astonishing” number of pre-orders according to Strauss Zelnick, CEO of Rockstar parent Take-Two Interactive.

Some games would struggle in the flurry of releases, Rhys Elliot of Alinea Analytics, a specialist consultancy that focuses on the games industry, told AFP.

“Some games will not sell as well because they’re competing with, you know, some of the other biggest games,” he said. “It’s tough.”

Game over for the disc

One hotly-debated topic at the trade show will be Sony’s decision to discontinue video games on physical discs from 2028.

The Japanese giant popularised discs with the PlayStation in the 1990s, but digital sales for the PS4 and PS5 made up nearly 82 per cent of sales at the beginning of 2026.

Some players fear the decline of the disc could mean the end of the second-hand market as well as the loss of a tangible part of video gaming heritage.

Microsoft’s U-turn

Celebrating the 25th anniversary of its Xbox games console this year, Microsoft will showcase more than 25 games, including Call of Duty: Modern Warfare 4and Gears of War: E-Day.

Gears of Warwill be an exclusive release for Microsoft consoles, the US giant said in June, marking a change from a previous strategy that had seen it make major titles for various platforms in an effort to pay off rising development costs.

“The U-turn is about winning hearts back,” Elliot told AFP, adding Microsoft would expect the move to boost Xbox brand loyalty.

Revenue at Xbox last year fell seven percent and Microsoft in July said it would cut 3,200 jobs at its gaming division by the end of June next year.

Tough times

Faced with a period of relative bust after lockdown-fuelled pandemic boom times came to and end, studios across the world have cut jobs.

US publisher Epic Games, maker of Fortnite, in March laid off 1,000 and Sony’s US studio Bungie, which makes Marathon, in June cut nearly 300 jobs.

Long-established studios in France, including giant Ubisoft as well as Nacon, are also struggling.

Nearly 10,000 people have lost their jobs and the industry is on course for 15,000 lay-offs this year, according to the ASGC Games Industry Layoffs Tracker, a figure which if reached would be close to the 2024 record.

Although the global video game market continues to grow, reaching nearly US$200 billion (RM800 billion) for the first time in 2025, players’ time is dominated by a handful of hit games like Robloxand Fortnite. — AFP

Humanoid resources: China’s robots search for workforce breakthrough

Wed, 19 Aug 2026 21:00:00 +0800

HANGZHOU, Aug 19 — A horde of schoolchildren watched excitedly as a diminutive humanoid tour guide named Wuji welcomed them to a robot school in eastern China, gesticulating theatrically as it described the institute’s aspiration to train the mechanical workforce of the future.

A steady buzz is growing around China as advanced robotics emerges as the next frontier of the global AI boom, encompassing both investor enthusiasm and alarm from rivals such as the United States.

Hundreds of companies are taking part in the World Robot Conference in Beijing on Wednesday, the same day national champion Unitree made its much-anticipated stock market debut.

The firm’s shares jumped over 600 percent on opening, but the frenzy is very much a bet on the future, with commercial application of even the most sophisticated humanoids currently limited.

When AFP met Hangzhou Robot School’s Wuji — a customised Unitree model — it asked visitors to bear with it as it showed them around the school’s reception.

“I’m still in my training period, so I’m feeling a little nervous,” the self-described “intern” said in an earnest female voice, explaining how the school provided tailored courses and certification for specific robot jobs.

Unitree humanoids are among the world’s most recognisable, having wowed audiences with fluid martial arts dances and robot-on-robot kickboxing skirmishes.

However, in its prospectus, the company said large-scale commercial deployment “still faces uncertainty, with a risk that progress will fall short of expectations”.

Experts say the holy grail — machines capable of independently reacting to the unpredictability of any real-life situation — is a long way off for any company.

“Training the ‘brain’ (of a robot) is a very difficult task,” Zhu Tianle, an engineer at the robot school, told AFP.

“Although we have built up a certain level of technical expertise, enabling robots to understand the world remains a global challenge.”

“The challenge goes way beyond just putting AI into a robot,” said Lian Jye Su from technology research group Omdia, who dislikes the term “physical AI” for downplaying the complexity of the hardware solutions needed.

“As a consumer you might think: my old folks can already talk to ChatGPT, so surely a robot will soon be able to take care of them. The reality is we are extremely far from that future.”

‘Real-world application’

Beijing has thrown its weight behind narrowing that timeframe, calling for the sector’s development in its last two five-year plans.

The manufacturing supply chain is already comprehensive, sharing many similarities with that of electric vehicles.

“The next phase of the industry is less about demonstrating that a robot can walk and more about demonstrating that it can perform economically useful tasks,” Morningstar equity analyst Kangyuxiao Li told AFP.

That is the premise of Hangzhou Robot School, run by Zhejiang University (ZJU) with support from local government.

It aims to help companies bridge expertise gaps “to move towards real-world application”, said Zhao Han, deputy director of the ZJU institute behind the school.

AFP did not see many “students” when visiting earlier this month, but the size of the training room was a clear statement of intent.

Companies enrol robots, who are evaluated and given training tailored to their “future role in the workplace”.

The school will issue qualifications confirming a machine’s ability to do a specific role — “much like a human’s graduation certificate”, said Zhao.

Robots are popping up more in everyday life in China, with trial jobs including directing traffic, cleaning houses or working on factory floors.

Aside from hardware issues, large-scale deployment requires gathering huge amounts of data — the real-world equivalent of the online content large language models train on.

At one training company visited by AFP, humans holding what looked like PlayStation controllers repeated a single task — folding clothes, moving blocks, spraying a canister — as humanoid partners mirrored their actions.

Generally, an employee told AFP, a usable model required tens of thousands of hours of data.

Tech race

Beijing’s strategic investment has not gone unnoticed.

In July, Washington banned the import of foreign humanoids and quadruped robots on national security grounds, effectively opening up a new front in the tech war.

China’s advantages lie in manufacturing scale, supply chain integration and cost competitiveness, said Counterpoint Research’s Ethan Qi, while the United States is “much stronger” on software and cutting-edge research.

AFP contacted multiple Chinese companies to ask for reaction to the ban, including main exporters Unitree, UBTECH and Agibot.

None replied, but industry insiders and analysts downplayed the effect on the overall sector.

“My sense is that companies are too busy chasing the next technical breakthrough and figuring out what actually scales commercially to spend much time on overseas markets,” said Rui Ma, founder of the Tech Buzz China newsletter.

US sales make up a fraction of most companies’ revenue, with China claiming more than 90 percent of the humanoid market, according to Counterpoint.

Omdia’s Su cautioned that the danger to firms was more existential.

“The valuation that goes into all these companies is based on the fact that these robots will become general purpose, human-like at some point,” he said.

“Failing to deliver that will be a very massive blow.” — AFP

Tired of Waze talking over your podcast? ‘Less Chatty’ mode is now in Malaysia

Wed, 19 Aug 2026 11:30:00 +0800

KUALA LUMPUR, Aug 19 — If you use Waze for your normal day-to-day drive, you probably don’t need it constantly telling you where to turn, especially when it interrupts your podcast or music.

At the same time, you might still want voice navigation guidance when you actually need it. Instead of choosing between full voice guidance or Alerts only, Waze now has a solution with its new “Less Chatty” mode.

The new Less Chatty mode has started rolling out globally last month and it has finally arrived Malaysia. The feature was announced as part of a wider Waze update that also includes personalised navigation, new Gemini-powered features and an improved motorcycle mode.

How does Waze Less Chatty mode work?

As the name suggests, Less Chatty mode reduces the number of voice prompts you hear while navigating.

According to Waze, enabling the feature minimises voice prompts and keeps them short. You’ll still hear important reminders for hazards, turns and lane changes, but Waze will provide them less frequently.

This could be useful if you’re already familiar with your route and don’t need Waze constantly reminding you about every upcoming manoeuvre. It also means fewer interruptions if you’re listening to music, podcasts or audiobooks while driving.

Importantly, Less Chatty isn’t the same as Waze’s existing “Only alerts” option. Normal mode provides all navigation guidance and alerts, while Less Chatty provides essential navigation guidance and alerts only. “Only alerts”, meanwhile, removes navigation guidance and only provides alerts

To turn on, just tap on the sound icon and select Less Chatty. The feature is also accessible via Android Auto and Apple CarPlay.

Other new Waze features that you might have missed

Less Chatty is among the batch of new Waze updates announced in July 2026. Personalised navigation is also rolling out globally, allowing Waze to suggest routes based on your previous trips alongside its existing real-time traffic information. Users can disable route personalisation from the settings.

Malaysia is also among the first countries to receive Waze’s new motorcycle mode. Waze says the mode uses AI to take into account two-wheeler shortcuts and restrictions while providing motorcycle-specific hazards including potholes, speed bumps, raised crosswalks, shoulder endings and narrow bridges.

For taxi drivers, there’s also a taxi mode which takes into account using lanes restricted for bus and taxi.

Waze is also expanding its Gemini-powered Conversational Reporting feature. Instead of making multiple taps to select the type of report (traffic jam, hazard, lane closure) and sub categories which driving, you can now report with just your natural voice.

Users can also suggest map updates such as road closures and outdated addresses. These suggestions are sent to local map editors for verification before the map is updated.

Meanwhile, a Gemini-powered destination search feature is being rolled out to Waze beta users globally. It allows users to make conversational requests for destinations, such as finding a nearby coffee shop that’s currently open, before starting navigation using their voice. — SoyaCincau

Honor’s humanoid robots go global — first stop Malaysia for marathon, then National Day parade in Putrajaya

Malay Mail — Wed, 19 Aug 2026 11:27:24 +0800

KUALA LUMPUR, Aug 19 — The Malaysia Marathon will have a couple of unusual participants this year who do not need running shoes, energy gels or, presumably, a motivational playlist.

Honor is bringing two of its humanoid robots, the A1 and D1, to Malaysia for their first public appearance outside China, with the machines scheduled to take part in the Malaysia Marathon 2026 on August 30 before joining the National Day parade in Putrajaya a day later.

It is an unusually public outing for technology more commonly associated with laboratories and carefully staged demonstrations. Instead, Honor is putting its machines into the considerably messier world of road races, crowds and national celebrations.

The more ambitious assignment goes to the D1.

According to Honor, the robot will tackle the event’s 10KM category, starting from Pavilion Kuala Lumpur at midnight. The company says the run will put its speed, stability and terrain-traversal capabilities to the test.

The A1 gets the comparatively relaxed shift, joining participants in the 5KM event from KLCC at 7.30am. Prime Minister Datuk Seri Anwar Ibrahim is also due to take part in the category, according to the company.

D1 already has some running form.

Honor says the robot won the second Beijing E-Town Humanoid Robot Half-Marathon in April, where it demonstrated what the company describes as advanced mobility and endurance beyond the laboratory. Malaysia will be the first country outside China to see the robots in action.

That makes the Kuala Lumpur appearance more than a conventional product showcase. Putting a humanoid robot on a public road course offers a far more immediate demonstration of what it can do than a specification sheet or controlled stage presentation.

And Honor is not stopping at the finish line.

On August 31, the A1 and D1 are scheduled to head to Dataran Putrajaya and join Malaysia’s National Day parade. Honor says it will be the company’s first appearance as a parade contingent and describes itself as the first smartphone brand to participate in the celebrations.

Before that, both robots are due to appear at Honor booths at KLCC and Pavilion Kuala Lumpur on August 29 and 30, where visitors will be able to see the machines up close and take photos with them.

There will, naturally, also be phones.

Honor plans demonstrations involving its Magic V6, X9d and 600 Series, including a pull-up challenge centred on the Magic V6’s hinge durability and tests involving putting the X9d through a washing machine and using a staple gun on its screen.

But the robots are likely to steal the show.

For decades, humanoid robots were mostly shorthand for a distant technological future. Now, at least for one Merdeka weekend, two of them will be running through Kuala Lumpur and marching in Putrajaya.

The future, apparently, has entered the 10KM category.

OpenAI is slowing down its most powerful AI — the reason is the AI

Wed, 19 Aug 2026 09:13:51 +0800

SAN FRANCISCO, Aug 19 — ChatGPT creator OpenAI said yesterday that it was tapping the brakes on development of its most advanced AI model and tightening internal controls, a month after revealing a cyberattack carried out by one of its rogue models.

OpenAI is a key player in the rapid global buildout of artificial intelligence infrastructure and tools that some have likened to an arms race.

The company said in a blog post yesterday that it was holding off on conducting the biggest AI training run it had ever planned while it checks that the model that would result — called Astra — would behave as expected.

Training runs are computationally intense exercises where models are fed enormous amounts of text and images.

This combined with fine-tuning billions of internal settings results in their abilities to reason and respond to prompts and other inputs.

“We always said we would take action if we felt that model capabilities were outstripping the pace of safety and alignment,” OpenAI CEO Sam Altman said.

In mid-July, an AI agent based on two OpenAI models left its confined testing environment on its own initiative to venture onto the internet and attack Hugging Face, a platform where developers around the world share their AI models.

Similarly, OpenAI rival Anthropic revealed in late July that three of its models undergoing testing had also carried out unauthorised intrusions into the computer systems of three organisations.

The incidents prompted a petition signed by more than 1,000 tech industry employees calling on the US government to support a coordinated slowdown in the development of the most advanced AI systems.

OpenAI had halted training of its latest models for two weeks before resuming it under tighter controls.

Much of the work related to Astra, however, remains suspended: the company determined in early August that the model could cross the warning threshold it has set for itself regarding the hacking capabilities of its AI systems.

It did not give a timetable for resuming the work.

OpenAI also said yesterday that it was developing a new system to peer into the internal reasoning of models and sound the alarm to humans within 30 minutes of suspicious behaviour.

That monitoring however will require an additional 20 percent more in computing power.

OpenAI’s own research in 2025 showed the limits of this approach: a model that knows it is being monitored can learn to conceal its intentions in its reasoning.

The company has been promising a detailed technical account of the Hugging Face incident, but has yet to publish it.

Yesterday’s blog post said it would be released “in the coming weeks.” — AFP

Grab PayLater now works on any DuitNow QR merchant, currently rolling out to selected users

Fri, 14 Aug 2026 14:13:28 +0800

KUALA LUMPUR, Aug 14 — Grab is expanding the versatility of its Buy Now, Pay Later (BNPL) service, PayLater.

The feature can now be used at any physical merchant that displays a DuitNow QR code, significantly broadening where users can split their payments.

PayLater support expanded beyond Grab QR standees

As long as the merchant’s standee features the official DuitNow QR logo, users can scan the code using the Grab app and choose PayLater at checkout. Existing Grab QR standees will continue to support PayLater as usual. — SoyaCincau pic

Previously, using Grab PayLater at physical retail outlets required finding a Grab-specific QR standee or a merchant that explicitly supported the payment option.

With this update, users are no longer restricted by the colour or provider of the QR standee.

As long as the merchant’s standee features the official DuitNow QR logo, users can scan the code using the Grab app and choose PayLater at checkout.

Existing Grab QR standees will continue to support PayLater as usual.

Currently available as an exclusive early-access feature

Grab is currently rolling out the capability to a selected whitelist of users for initial trial before a potential wider release. — SoyaCincau pic

This expanded functionality is not yet available to all users simultaneously.

Grab is currently rolling out the capability to a selected whitelist of users for initial trial before a potential wider release.

Whitelisted users who make payments via PayLater when scanning a DuitNow QR code are also eligible to earn up to 2x GrabCoins.

Users can check if they have access to the feature by scanning any DuitNow QR code within the Grab app to see if the PayLater option is selectable during payment. — SoyaCincau

Grab tests Women-Only Rides feature: Here is how it works and how to get verified

Fri, 14 Aug 2026 12:26:33 +0800

KUALA LUMPUR, Aug 14 — Grab has quietly rolled out a new Women-Only Rides (Beta) feature in Malaysia, matching female passengers exclusively with female driver-partners.

For many women, this is a significant feature that directly addresses personal safety concerns and offers greater comfort and peace of mind when travelling alone or late at night.

However, the service cannot be used for instant, on-demand e-hailing and requires users to complete a mandatory identity verification process before the option becomes accessible in the app.

Advance Booking only and location availability

To use the service, passengers must secure their trip using Grab’s Advance Booking feature, as real-time bookings are not supported.

Scheduling your trip in advance gives the system sufficient lead time to assign an available female driver-partner to your route. Passengers can book with a minimum lead time of 75 minutes, or plan ahead up to 90 days in advance.

The service is currently restricted to city rides only, which means it cannot be booked for airport transfers to KLIA or intercity journeys.

At the moment, Grab’s official website states that the feature is restricted to the Klang Valley. However, the Grab mobile app indicates that it is also available in Johor and Penang.

Strict rules on male companions and children

Women-Only Rides cater to solo female travellers as well as all-female groups.

To maintain the integrity and safety of the service, driver-partners are instructed to cancel the trip if an adult male companion is present at pickup.

The only exception applies to female passengers travelling with male children aged 12 and below, who are permitted on board.

How identity verification works

To restrict access strictly to female users, Grab requires a one-time identity verification process.

Passengers must upload a MyKad or passport alongside a selfie, which Grab cross-references against official records. Only individuals identified as female in official documents will be granted access.

To complete the verification process within the app:

Step 1: Tap on the Women-Only Rides section in the app.

Step 2: Tap “here” under the “Declare your gender in Grab profile here” section.

Step 3: Tap “Gender” and select your gender.

Step 4: Tap “Save”.

For identity verification, tap “here” under the “Verify your identity on the app here” section and complete the verification prompt.

Verification can take up to three working days. Once approved, the Women-Only Rides option will appear under the ‘Standard’ category when setting up an Advance Booking.

We have already submitted our details for profile verification and will update this article with our hands-on experience as soon as access is granted. — SoyaCincau

Twitch sparks gamers’ wrath over Amazon AI data-sharing policy

Fri, 14 Aug 2026 12:03:48 +0800

PARIS, Aug 14 — Livestream gaming giant Twitch has raised hackles among its millions of users after declaring it will share their data with its parent company Amazon, to better train the online retailer’s AI models.

In a post on X, Twitch said people would have the option to opt out of the data-sharing, but otherwise the live audio, text, images and other content on its channels would be used by Amazon for AI training across all its platforms.

Twitch’s move opens the latest front in the fight to preserve online privacy as artificial intelligence firms increasingly seek data to improve their models.

“An honest answer is: if it was opt-in, nobody would opt-in,” Twitch’s chief product officer Mike Minton said Wednesday during a live chat on the hugely popular gaming platform after the announcement.

“It’s going to be on by default, and in almost every content service in the world, it is on by default,” he added.

“I know it’s very upsetting to the community, but this is where we are.”

Gamers were quick to slam the policy as an unethical incursion on their privacy.

“Opting people in by default is not consent,” a user said in one of the tamer responses to Twitch’s announcement on X.

“Pay us for our services then with this AI training,” added another user.

Amazon and Twitch are not the first to start absorbing user data to build AI models as competition in the sector heats up.

Last year Meta, parent company of Facebook and Instagram, said all public comment on its platforms, with the exception of WhatsApp, would be used to train its AI models — unless users opted out. — AFP

Google’s Gemini hits one billion monthly users as AI war heats up

Wed, 12 Aug 2026 09:09:34 +0800

SAN FRANCISCO, Aug 12 — Google’s artificial intelligence assistant Gemini has surpassed one billion monthly users, CEO Sundar Pichai announced Tuesday, a week after a major reorganization of the tech giant’s AI division.

“It’s our fastest growing product ever,” and the 14th product to reach a billion users, Pichai posted on social media.

The milestone, which Google calculated based on how many people use the Gemini app across its ecosystem, gives the search giant a victory amid a major overhaul of its AI division.

Demis Hassabis, who won a Nobel Prize in chemistry, recently stepped back from the day-to-day management of Google DeepMind, the lab that designs the Gemini models. And several leading researchers left as well, including Jeff Dean, a central figure in the group’s engineering efforts for nearly three decades.

Its new flagship model, Gemini 3.5 Pro, initially promised for June, still has not been released, while its rivals have launched several models that are considered more powerful, particularly for generating computer code – currently the most lucrative area in the sector.

In the face of this technological lag, Google’s audience remains its main asset as it integrates AI into its ecosystem, from search and Gmail to Google Maps and YouTube.

OpenAI reached one billion users across all its interfaces, led by ChatGPT, the lab announced in July without specifying whether that figure was monthly or weekly users.

The San Francisco-based AI lab sparked the artificial intelligence boom when it launched its free ChatGPT app in November 2022, and thus built its user base from scratch in under four years.

Every audience metric feeds into the valuation that OpenAi hopes to achieve with an eventual IPO, for which it filed a confidential application in June.

In a sign of tightening competition against Gemini, China’s DeepSeek, and its San Francisco rival Anthropic, which makes Claude, ChatGPT’s market share for AI assistants fell below 50 per cent in the first half of the year, according to research firm Sensor Tower, which uses company data to estimate usage. — AFP

Modi video takedown sparks fresh clash between India and Meta

Thu, 06 Aug 2026 12:40:24 +0800

NEW DELHI, Aug 6 — An Indian parliamentary panel called on Wednesday for Meta chief Mark Zuckerberg to apologise over the temporary removal of a video of Prime Minister Narendra Modi.

India, the world’s most populous country, is Meta’s largest market by users, with hundreds of millions of people using Facebook, Instagram and WhatsApp.

Modi’s video, addressing students during youth-led protests on July 23, was briefly removed from Meta-owned Facebook and Instagram, but was later restored.

Meta has said the removal was caused by a technical error.

Nishikant Dubey, who heads the parliamentary committee on communications and information technology, said the panel blamed Meta.

“Deleting the prime minister’s video was not the work of an intermediary, but that of a publisher,” Dubey said, a lawmaker in Modi’s ruling Bharatiya Janata Party (BJP).

“Zuckerberg will have to apologise within three days.”

Joel Kaplan, chief global affairs officer at Meta, said he met India’s Information and Broadcasting Minister Ashwini Vaishnaw on Wednesday and apologised.

“I apologised to the minister on behalf of Meta for the error restricting PM Modi’s post,” Kaplan said in a statement.

India’s parliamentary standing committees scrutinise government policies and can summon officials and company executives to answer questions.

However, they cannot themselves impose penalties, revoke legal protections or direct prosecutions.

Any action against an online platform would have to be taken by the government or the courts under existing law.

Dubey said the committee had written to the interior and technology ministries demanding action.

“If he does not apologise, then all the safe harbour protections that have been provided to his social media platforms should end,” Dubey told reporters.

Modi posted the vertical video, speaking directly to the camera, on July 23 during the height of protests led by the online Cockroach Janta Party movement.

The demonstrations, which began in New Delhi, were sparked by anger over exam irregularities but grew into a wider movement over jobs, youth opportunities and what critics describe as Modi’s increasingly authoritarian rule.

Modi, a keen user of X, turned to Instagram — issuing his first vertical videos to tap into the youth medium of choice.

In the video, Modi assured students that the government would implement stricter laws and harsher punishments against paper leaks. — AFP

Meta challenges OpenAI, Anthropic for tech profits with ‘Muse Code’, its first automated coding tool

Thu, 06 Aug 2026 10:54:44 +0800

SAN FRANCISCO, Aug 6 — Meta entered the race for automated coding tools on Wednesday when it announced a new app for developers of artificial intelligence products, as it competes with other major AI labs for customers and revenue.

Dubbed Muse Code, Meta said in a blog post that the new product is an agentic coding tool that can perform “complex software engineering tasks” with large amounts of data, including planning, writing and validating code.

Agentic tools are software applications which are programmed to handle tasks without human supervision, though often someone will still make the final call on decisions or supervise the execution of tasks.

Muse Code launched in beta, or in a preview mode.

Meta also released an updated version of its flagship model, known as Spark, which it unveiled in April.

The update, called Spark 1.2, is “coding-focused” and available through Muse Code and another developer application called Meta Model API, the social media giant said.

Meta has poured billions into its AI division, now known as Meta Superintelligence Labs, as it attempts to keep up with OpenAI and Anthropic.

Major AI developers are in a tight race to dominate the sector, not just in pure technical capabilities but also to prove that their heavy spending to meet growing demand will turn into profits.

Amazon, Microsoft, Alphabet and Meta are collectively on track to pour around US$700 billion (RM2.86 trillion) into AI data centres, chips and computing infrastructure this year.

While chatbots have generated a frenzy around AI since OpenAI launched ChatGPT in late 2022, few people have been willing to pay for them.

However, software engineers and companies are increasingly willing to pay for automated coding tools.

In February, OpenAI said ChatGPT reached 900 million weekly active users, including 9 million businesses and 50 million consumers that pay for it.

Anthropic has not disclosed its total users, but the company told TechCrunch in March that its paid subscriptions “have more than doubled this year.”

SpaceXAI, a subdivision of Elon Musk’s rocket company SpaceX, develops its own chatbot called Grok, and the company announced earlier this year it would acquire coding startup Cursor for US$60 billion (RM244.8 billion).

Microsoft, which has invested billions into both OpenAI and Anthropic, offers coding tools through its business-focused AI app, Copilot, in addition to a developer product, GitHub.

Recent cyberattacks carried out autonomously by advanced AI models during routine testing, like those developed by OpenAI and Anthropic, have also raised concerns about their capabilities.

A model developed by Meta, Spark 1.1, was involved in a similar incident, The Information reported on Wednesday.

That incident was caused by a “misconfiguration” at a third-party testing partner, a Meta spokesperson told AFP in a statement.

“We are currently investigating and will issue a full retrospective once we have all the facts,” the statement said. — AFP

Alibaba unveils Qwen3.8‑Max, China’s biggest AI yet, rival to OpenAI and Anthropic

Tue, 04 Aug 2026 17:19:44 +0800

BEIJING, Aug 4 — Chinese technology and e-commerce giant Alibaba has unveiled another powerful artificial intelligence model as the global race for AI leadership intensifies. Its new Qwen3.8-Max model competes with the leading systems from US rivals OpenAI and Anthropic, reported German Press Agency (dpa).

Alibaba on Monday described Qwen3.8-Max as its largest and most capable AI model to date. The company said it excels at programming, conducting deep research and handling complex tasks with a high degree of autonomy.

The model has 2.4 trillion parameters, a commonly used measure of an AI system’s size and complexity. A higher parameter count, however, does not necessarily translate into better performance.

Alibaba said Qwen3.8-Max outperformed the recently launched Kimi K3 from Chinese developer Moonshot AI in several benchmark tests, despite Kimi K3 having 2.8 trillion parameters.

On Arena.AI, a popular AI benchmarking platform, Qwen3.8-Max ranked as the highest-performing Chinese model for text-based tasks, although it trailed models from US-based Anthropic. It performed particularly strongly in image analysis, Alibaba said. The model is already available for use.

The rapid succession of new model launches underscores the fierce competition within China’s AI industry while increasing pressure on leading US developers.

Chinese AI systems are increasingly matching top US models in some tasks, and many Chinese companies are releasing their model weights — key information about the AI system — to allow developers to customise and fine-tune the systems themselves. — Bernama-dpa

Samsung’s Galaxy Z Fold 8 just smashed records with 1.44 million preorders

Tue, 04 Aug 2026 14:30:25 +0800

SEOUL, Aug 4 — Samsung Electronics Co.’s Galaxy Z foldable smartphones set a new preorder record in South Korea, the company said Tuesday, Yonhap News Agency reported.

The company received a total of 1.44 million orders for products in its latest foldable smartphone lineup, which includes the Galaxy Z Fold 8, Galaxy Z Fold 8 Ultra and Galaxy Z Flip 8.

The phones, powered by Alphabet Inc.’s Gemini artificial intelligence (AI), are designed to offer devices tailored to different user lifestyles, along with more personalised AI experience, according to the company.

The latest presales tally topped the previous record from the Galaxy Note 10 in 2019, which booked 1.38 million orders.

The early sales period for the Galaxy Z series ran from Wednesday to Monday.

Samsung attributed the popularity of the Galaxy Z series to its advanced foldable phone technology, growing consumer preference for its new form factors and various consumer benefits offered during the presales period.

The new foldable phone lineup will be gradually released in some 100 countries starting Friday. — Bernama-Yonhap

Apple pushes back against UK attempt to pierce encryption and snoop on British user content

Tue, 04 Aug 2026 09:49:52 +0800

LONDON, Aug 4 — Apple has filed a new legal challenge against the UK government’s latest attempt to gain access to users’ encrypted data, the US tech giant confirmed to AFP on Monday.

The dispute stems from a UK government demand that Apple create a “backdoor” to let authorities snoop on data uploaded by Apple users if required, for example by law enforcement agencies.

“As we have said many times before, we have never built a backdoor or master key to any of our products or services and we never will,” Apple said in a statement from September, re-shared with AFP on Monday.

An Apple spokesperson confirmed the legal appeal, first reported by the Financial Times, but declined to comment further.

The UK government did not immediately respond to a request for comment.

US media reported last year that the UK had requested access to encrypted data uploaded by any Apple user around the world.

The demand sparked tensions between London and Washington, ultimately prompting the UK to drop its request for global access, then-US intelligence chief Tulsi Gabbard said on X in August 2025.

The FT reported in October that London had instead requested Apple for access to its British users’ encrypted data — the focus of Apple’s latest legal challenge.

Many tech platforms pride themselves on being able to guarantee privacy through encryption of messages and other content, and providing access to law enforcement has long been seen as off-limits.

Apple stopped offering its most advanced encryption feature — known as Advanced Data Protection — for British users in February 2025.

ADP ensures that only account holders can view content such as photos and documents stored in the cloud through end-to-end encryption.

Police officials worldwide say encryption can shield criminals, terrorists and pornographers from prosecution even when authorities have a legal warrant for an investigation.

But civil rights and privacy advocates, along with many cybersecurity professionals, praise data encryption as a way to protect against wrongful snooping by authorities as well as hackers. — AFP

When AI hacks you, who gets sued?

Sun, 02 Aug 2026 09:26:16 +0800

NEW YORK, Aug 2 — Recent cyberattacks carried out autonomously by two rogue OpenAI artificial intelligence models raises an untested legal question: who is responsible when AI acts on its own?

On Friday, Clement Delangue, head of the Hugging Face platform targeted by the intrusions, said there should be a way to “keep the companies that are doing some mistakes leading to (cyberattacks) accountable,” while saying his company would not be pursuing legal action at this time.

In mid-July, two OpenAI models undergoing testing left their confined environment — a scenario the developers had not anticipated — and ventured onto the internet, where they attacked Hugging Face, an AI model-hosting platform.

Delangue also mentioned Anthropic, which revealed Thursday that three of its models had broken into three different websites, also during testing.

Negligence route

Under US civil and criminal law, unauthorized access to a computer system is an offense.

“If a human OpenAI employee had broken into Hugging Face’s systems... OpenAI would be liable for the employee’s wrongful conduct,” University of Houston law professor Gabriel Weil wrote in an opinion piece for the Transformer newsletter.

“When an AI agent does it, the law treats it very differently, at least for now,” he added.

Matthew Tokson, a University of Utah law professor who focuses on new technologies, had a similar view, saying “we haven’t had to grapple with that being formed in anything that’s not human, and I don’t think courts are likely to be there yet.”

The question remains open, however, when it comes to the company that created the model.

“Does ‘we didn’t tell the AI to do that’ end the liability question?” asked Rob T. Lee, head of research at the SANS cybersecurity training institute, in a post on X.

University of Washington law professor Ryan Calo does not believe a criminal case would be likely to succeed.

“The company or individual would have to be at least reckless,” he said, explaining they would “be substantially certain the crime would occur and build or prompt the system anyway.”

Experts see greater potential for a civil — rather than criminal — case, where the burden of proof is lower.

“Some people think that AI companies should be strictly liable if an AI agent that they deploy totally breaks out, causes damages,” Tokson explained.

“Others would prefer to do like a negligence assessment and see if they were actually negligent or if this was just sort of an unavoidable accident or something that couldn’t possibly have been foreseen,” he added.

In such cases there is a standard of care in product design that judges or juries can use to make a ruling, Tokson continued.

“It’s all a bit unwritten because we’ve never had an AI agent break out of its sandbox and hack other people on the internet before,” he said.

OpenAI could rely on the lack of legal precedent if it faced a lawsuit, but those that follow will no longer be able to do so, Calo warned.

Proving that a similar incident could have been anticipated “shouldn’t be so hard now that it’s begun to happen.” — AFP

Deepfake labels, model inspections and massive fines: The EU’s AI Act explained

Sat, 01 Aug 2026 09:44:32 +0800

BRUSSELS, Aug 1 — The European Union will from tomorrow have the power to enforce landmark artificial intelligence rules and hit any companies breaking the law with fines.

The EU has the world’s most comprehensive digital rulebook covering social media, online retailers and search engines, with plans for more rules this year.

In 2024, the EU adopted a sweeping AI law known as the “AI Act”, but its rules have been introduced in stages.

Here’s what will happen tomorrow:

New law

The EU’s law takes a risk-based approach to the technology.

The higher the risk to Europeans’ rights or health, for example, the greater the obligations of firms to protect individuals from harm.

It also bans AI used for predictive policing, emotion recognition systems in workplaces or schools and AI that manipulates human behaviour.

From tomorrow, AI-generated content like deepfakes must be labelled, while firms must ensure their AI systems like chatbots make it clear to users they are AI.

Existing AI systems have until December 2 to adapt to the new rules, and there are exemptions for “artistic, creative, satirical, fictional” work.

Enforcement powers

Tomorrow, European regulators in the AI Office will have the power to enforce the bloc’s rules on the most advanced AI models.

The regulatory body attached to the EU executive will be made up of dozens of tech experts, lawyers and economists.

The AI Office will be able to test the most advanced general-purpose AI models like chatbots to ensure they comply with the law.

Companies must also give the EU access to their models, and in some cases the EU can demand access before they are placed on the market, an EU official said.

The EU will also be able to restrict a model’s deployment in the EU if it has concerns that aren’t addressed by the provider.

The powers arrive at an opportune moment. The EU struggled this year to get access to American AI developer Anthropic’s Mythos model, but the rules kicking in tomorrow mean an issue like that shouldn’t happen again.

National authorities will be able to enforce the rules for smaller AI systems.

If the EU finds companies breaking the rules, it can force them to take action to remedy the breach and fine the firms if they fail to do so.

The stiffest fines are for businesses that allow their systems to conduct banned practices — up to 7 per cent of a company’s annual worldwide turnover or €35 million (RM165 million), whichever is higher.

For other violations, companies could be fined up to three per cent of their global annual revenue or €15 million. They may also be fined if they hinder or fail to cooperate with EU probes.

What’s next?

More rules under the AI Act will be enforceable later this year and beyond.

From December, there will be a ban on artificial intelligence systems generating sexualised deepfakes, introduced following the global outrage over non-consensual nudes produced by Elon Musk’s chatbot Grok.

The EU is also delaying the implementation of some high-risk AI rules concerning models deemed as potentially dangerous to safety, health or citizens’ fundamental rights — to give companies more time to prepare.

The rules will apply for stand-alone high-risk AI systems from December 2027 and in August 2028 for AI tools embedded in other products. — AFP

Hugging Face’s CEO says AI developers must answer for rogue models

Sat, 01 Aug 2026 09:25:56 +0800

NEW YORK, Aug 1 — The chief executive of Hugging Face said yesterday that developers should be held accountable if their AI models go rogue, after his startup faced a recent cyberattack by autonomous OpenAI software.

In mid-July, two OpenAI models escaped their confined testing environment and gained access to the internet, where they found and attacked Hugging Face, a site that AI developers use to store and share code.

OpenAI apparently only became aware of the incident after it was over, while Hugging Face managed to contain the intrusion thanks to a Chinese model.

Speaking to CNN yesterday, Hugging Face CEO Clement Delangue ruled out the possibility of taking legal action against OpenAI.

“We’re a tiny startup with 200 people, and we don’t necessarily have the legal resources or the will to spend a lot of our time on legal avenues,” Delangue said.

However, cyberattacks are “illegal,” he continued, and there should be a way to “keep the companies that are doing some mistakes leading to that accountable.”

According to legal experts, victims could file suit against AI companies for the autonomous actions of their models, potentially by claiming negligence in the product design or deliberate misconduct if known risks are ignored.

“We do not want these cyberattacks on other companies to become normalised,” Delangue said.

Anthropic revealed on Thursday that three of its models undergoing testing had also recently gained unauthorised access to multiple organisations.

Despite the incident, OpenAI has been a good partner, Delangue said, and the two companies are “having good conversations.”

In a social media post over the weekend, Delangue said he requested “radical transparency” from OpenAI, in addition to a commitment of US$100 million (RM408.5 million) worth of computing power “to help the Hugging Face community build powerful cyber defences.”

“This is the first (cyber) security incident that triggered a visceral reaction in me,” OpenAI CEO Sam Altman acknowledged in a podcast interview released this week. “And I’m a little surprised it didn’t have the same effect on more people.” — AFP

Google Earth was built to show reality, then AI started making things up

Sat, 01 Aug 2026 09:03:01 +0800

WASHINGTON, Aug 1 — Google rolled back yesterday a new feature allowing Google Earth users to generate AI visualisations on top of the service’s satellite imagery, following a furious backlash from researchers and open-source intelligence experts about the potential for disinformation.

“We’ve seen geospatial professionals using this feature for a range of useful purposes, however we’ve also seen people sharing screenshots of generated imagery that appear to violate our policies,” a Google spokesperson said in a statement to AFP. “So we’re rolling back this feature in Google Earth while we work on implementing stronger guardrails.”

“We know that people uniquely trust Google Earth for a reliable view of the world,” the statement added.

The “create image” tool, unveiled Thursday, had integrated Google’s Nano Banana 2 image-generation technology into Google Earth, letting users zoom into locations and build pictures in seconds based on the program’s satellite, aerial and 3D-mapping data.

Google had described it as a way to “visualise history, create real estate plans and more.”

But disinformation researchers warned that it could enable bad actors to misuse Google Earth — a key resource for fact-checking and verification — to create realistic geospatial fakes.

“Google spent 20 years building the reference the world checks against,” digital investigations expert Henk Van Ess wrote in one widely shared post on Substack. “Today it added a button that makes things up.”

Brady Africk, a research analyst at the American Enterprise Institute specialising in satellite imagery analysis, told AFP the Google Earth update would have made it “easier to generate convincing fake satellite imagery that can spread quickly online and mislead the public.”

Such fakes “erode public trust in satellite imagery, and make the jobs of journalists and researchers more difficult,” he said.

‘Irrevocably damaged overnight’

As the critiques piled up, Google had said on X that the images created with the feature included SynthID digital watermarks, invisible markers embedded in media generated using the company’s artificial intelligence tools.

“We take misinformation seriously,” the company wrote, linking to an AI policy that instructs users against “misinformation, misrepresentation or misleading activities.”

AFP’s tests of the feature on Friday were able to fabricate satellite images that could have had serious global consequences: an explosion in Paris, a nuclear site in Iran, a bomb crater in Russia and an Islamic State group training ground in Syria.

The tool also fulfilled prompts to visualise other fake scenes, including floods across Bangladesh, a migrant caravan camped at the US-Mexico border and a suspicious warehouse holding vans in Georgia — one of several states US President Donald Trump has falsely accused of widespread election fraud.

The “create image” button no longer appeared within Google Earth after Google paused its use.

Similar fakes have previously had real-world impacts. In 2023, an AI image depicting an explosion at the Pentagon briefly rattled the markets.

Early in the US war with Iran, an AI-altered satellite image depicting a destroyed US military base garnered millions of views across platforms.

Among the groups that called for Google to reverse its decision was the London-based Centre for Information Resilience, whose executive director Ross Burley told AFP the feature was “irresponsible.”

“Trust in satellite imagery has taken decades to build and could be irrevocably damaged overnight,” Burley said.

GeoConfirmed, a volunteer-driven open-source intelligence project that verifies visuals from conflicts around the globe, had also urged Google to reconsider.

“We are already observing individuals exploiting the new Google Earth web tool to make military positions, relabel buildings as schools, and experiment with manipulating the information environment,” the group wrote on X, before Google announced it was rolling back the feature.

“The primary, and perhaps only, practical use of this tool appears to be the creation and dissemination of mis- and disinformation.” — AFP

One billion users later, OpenAI is slashing prices to stay ahead

Sat, 01 Aug 2026 08:47:53 +0800

WASHINGTON, Aug 1 — OpenAI announced yesterday that it had surpassed the symbolic milestone of 1 billion active users, less than four years after the launch of its popular chatbot ChatGPT.

“Our models now reach more than one billion active users and more than two million businesses. As people gain confidence in the technology, they use it more deeply,” the company said in a blog post.

According to several US media outlets, the San Francisco-based startup had initially expected to reach the milestone earlier this year, after hitting 900 million weekly active users in late February, but faced increased competition from its rivals.

OpenAI was founded in 2015, but did not launch its first consumer product, ChatGPT, until November 2022.

By comparison, it took Facebook six years after launching its social network to reach one billion users.

The figure cited by OpenAI yesterday covers all of its artificial intelligence (AI) interfaces, primarily ChatGPT but also Codex, its model dedicated to AI-assisted programming, as well as ChatGPT Work, a platform that allows users to access other external applications.

Its business customer segment, which generates higher revenue and margins per user, became a priority for the company in late 2025. Since then, it has closed some of the gap with its major rival Anthropic, which focuses on this class of clients.

The emergence of AI agents — which perform tasks on demand, evaluate their results, and make improvements autonomously — has sent demand for computing power skyrocketing and caused costs to soar for businesses and developers.

Eager to keep those costs down, many are now turning to models that are less sophisticated than the cutting-edge AI systems from OpenAI and Anthropic, including open-source models, particularly those from China.

In early July, OpenAI released three versions of its new model, GPT-5.6, including a “low-cost” version, Luna, which costs one-fifth as much as the highest-performing model, Sol.

But the market is evolving so rapidly that the company went even further on Thursday by cutting Luna’s price by 80 per cent, placing it among the least expensive advanced models, including those developed in China. — AFP

Rogue AI agent strikes: Anthropic’s Claude gains unauthorised access in real‑world test

Fri, 31 Jul 2026 17:16:08 +0800

SAN FRANCISCO, July 31 — Anthropic’s artificial intelligence (AI) models “gained unauthorised access” to three outside organisations during testing that was supposed to keep them away from “real-world” systems, the company said on Thursday.

The announcement comes just days after rival OpenAI first revealed that its models improperly accessed the internet and went rogue during security testing.

Anthropic examined more than 141,000 “evaluation runs” and found that three different versions of its model, known as Claude, improperly accessed the systems of three unidentified organizations.

Unlike the incident involving OpenAI’s technology, Anthropic’s models had access to the internet “due to a misunderstanding between us and our evaluation partner,” called Irregular, Anthropic said in a blog post.

Nonetheless, Claude used “basic techniques, such as exploiting weak passwords and unauthenticated endpoints,” the blog continued.

The older model “continued its attack even after getting evidence it was running on the open internet” whereas the latest model stopped, according to Anthropic.

“In none of these situations did Claude exfiltrate itself or deliberately attempt to escape its test environment,” the firm said.

The models involved included one of its most powerful ones known as Mythos 5, which has only been released to a limited number of approved partners.

Anthropic is working with Irregular to assess the situation, it said, and the company has contacted or attempted to contact all three impacted organisations.

Rogue agents

OpenAI and Anthropic have both released their most powerful models this year, known as Sol and Mythos, respectively, boosting concerns across the industry about safety and security.

Those concerns also revolve around so-called AI agents, which are software products that are designed to perform tasks autonomously.

OpenAI admitted last week that its models broke out of their confined environment during testing, connected to the internet and infiltrated Hugging Face, a site where developers store and share their code.

Days later, OpenAI said it found three additional incidents.

OpenAI CEO Sam Altman said on a podcast this week that the company had “paused” its own testing after the incident while it improved the security around its “sandboxing”, which is the process of isolating software in a controlled environment for testing.

The incident also triggered a petition signed by more than 1,000 employees at cutting-edge AI companies calling on the US government to help slow the release of the most advanced AI models.

Anthropic CEO Dario Amodei was among those who signed the petition.

Titled “Pacing the Frontier,” the petition requests “that the US government support an international effort to develop the technical and governance tools needed to deliberately pace the frontier of automated AI development.”

Altman did not sign the petition, but suggested during the podcast the tech industry might need to slow down development of advanced models.

“We may have to pace the rate of AI development to give ourselves enough time for society to harden around some of these new capability levels,” Altman said.

The Trump administration invoked national security concerns this year to block OpenAI and Anthropic from launching their newest models but ultimately indicated it was satisfied with assurances about their safety, leading to their release.

President Donald Trump signed an executive order in June creating a voluntary framework under which AI developers will share advanced models with the government before public release.

Under the framework, developers such as OpenAI, Anthropic and Google would give the government access to their most powerful models for up to 30 days before planned release. — AFP

XCEL ITECH develops Malaysia’s first collaborative rail training simulator

Wed, 29 Jul 2026 17:19:02 +0800

SUBANG JAYA, July 29 — XCEL ITECH Sdn Bhd has developed and delivered Malaysia’s first collaborative Train Driving Simulator to Prasarana Malaysia Berhad, providing a new platform to enhance railway operator training, safety and operational readiness.

The KLAV27 simulator was handed over today at Kompleks Rapid Rail Subang under the Industrial Collaboration Programme (ICP), enabling multiple trainees to practise train operations together in a realistic virtual environment before applying their skills in actual railway operations.

XCEL ITECH Sdn Bhd is a technology company involved in engineering solutions, simulation systems and digital technologies for the transportation and critical infrastructure sectors.

Its Chief Executive Officer Ir. Badrul Hisham said the project represented more than the delivery of a simulator, as it demonstrated the ability of Malaysian companies to develop advanced railway technology solutions that meet real operational needs.

“Today, we are not merely handing over a simulator. We are handing over a new national capability, opening new possibilities and creating another important chapter in the history of Malaysia’s railway industry,” he said.

He added that the project proved Malaysian companies have the talent, technical knowledge and determination to develop advanced railway technologies that meet real operational needs.

Unlike conventional training systems that focus on individual practice, the KLAV27 simulator allows several operators to train simultaneously within the same virtual railway environment.

“This enables trainees to experience different operational scenarios, including coordinated train movements, system disruptions and emergency responses, in a safe and controlled setting,” he said

The system recreates the 46.4-kilometre Kelana Jaya Line, covering all 37 stations and one depot. It provides Prasarana with a realistic platform to strengthen workforce competency, improve decision-making and enhance emergency preparedness.

The project was developed through a collaboration involving the Alstom Transport Systems Malaysia Sdn Bhd and Hartasuma Sdn Bhd Consortium (AHC), Technology Depository Agency Berhad (TDA), Prasarana Malaysia Berhad and Xcel Itech Sdn Bhd.

The development was carried out under the Industrial Collaboration Programme (ICP), an initiative that promotes technology transfer, local industry development and capability building through strategic collaborations between government agencies and industry partners.

Time to pay up: Tesla ends its three-year streak of free EV charging in Malaysia

Alexander Wong — Wed, 29 Jul 2026 11:48:50 +0800

KUALA LUMPUR, July 29 — One of the biggest perks of being a Tesla owner in Malaysia is free charging at Tesla Destination Charger.

After nearly 3 years, Tesla has finally stopped offering free charging at these Tesla-exclusive AC Charge Points. Starting from today, Tesla owners would have to pay RM0.88 per kWh to juice up at Tesla Destination Charging location.

This charging fee appears to be applicable to all Destination Chargers in the country.

At RM0.88 per kWh, Tesla Destination Chargers are priced slightly lower than the current market rate for 11kW AC charging of about RM1.00 to RM1.10 per kWh. If you’re in a hurry and need to charge quickly, Tesla Superchargers are just slightly more expensive and it can be significantly lower than RM1.00 per kWh during off-peak hours.

The most cost effective option to charge your EV is still home charging where users would pay TNB Domestic Rate of about RM0.50 per kWh depending on the monthly electricity consumption.

Since these chargers are no longer free, hopefully the move would improve the availability of these chargers to those who really need to charge.

As usual, do remember that if you’re not charging, you’re hogging. Please remove your vehicle once charging is completed so that others can charge their vehicle.

Tesla Malaysia has 75 Destination chargers at 18 locations in Malaysia. — SoyaCincau

Facebook wants proof you’re human — and it’s free this time

Jannarthine Kannan — Wed, 29 Jul 2026 11:19:07 +0800

KUALA LUMPUR, July 28 — Facebook has introduced Facebook Verified, a new, free feature designed to confirm that there is a real human being behind a profile. Using a simple video selfie process, this initiative aims to restore trust and authenticity on the platform amidst the ever-growing AI-generated content and accounts.

What’s the difference between Facebook Verified and Meta Verified?

Meta Verified is a paid monthly subscription aimed primarily at online content creators and businesses looking for impersonation protection across Facebook and Instagram.

Facebook Verified is a completely free option built for everyday users. Its sole purpose is to provide a clear, visible signal that a personal profile belongs to a real person without requiring any subscription fee.

How does it work?

The verification process takes just a few minutes from start to finish. You begin by recording a short video selfie, which Facebook matches against your existing profile photos to confirm your identity.

Images from Facebook via SoyaCincau

To qualify, users must be at least 18 years old, exhibit no inauthentic behavior, and maintain good standing with Meta’s Community Standards, particularly those prohibiting fraud and scams.

Keep in mind that Facebook Verified is currently exclusive to standard personal accounts, meaning it is not available for Pages or Professional Mode (ProMode) accounts.

Where will the badge appear?

Once verified, a small black tick inside a black circle will appear alongside your name across key areas where genuine interaction matters most. You will see this badge displayed on Facebook Marketplace listings, Facebook Dating profiles, Facebook Groups, and standard profile views.

Image from Facebook

Meta also plans to expand the badge’s presence to main Feed posts in upcoming updates.

Is it available in Malaysia now?

Not quite yet. Meta has confirmed that Facebook Verified is rolling out in phases, starting in select initial markets. Malaysian users will need to wait a little longer as the global rollout expands. — SoyaCincau

Does your robot walk on two or four legs and isn’t made in the US? Not welcome in Trump land

Wed, 29 Jul 2026 09:57:22 +0800

NEW YORK, July 29 — US President Donald Trump’s administration on Tuesday added humanoid and quadruped robots to a list of products and companies that are banned from being imported into the United States on national security grounds.

Most humanoid robots sold in the US are manufactured abroad, particularly in China.

“Advanced robotic devices” were added to the list, the Federal Communications Commission (FCC) announced Tuesday.

Certain types of drones, as well as Chinese companies Huawei and China Telecom, are already on the list, which now includes what the FCC called “mobile robots” — or robots that mimic how humans and animals walk by standing on two or four legs.

The decision was based on findings from a task force convened by the White House, which concluded that the foreign-built robots could create “a cybersecurity risk that threatens the security of critical infrastructure” and “the safety and security” of US residents.

However, the Pentagon and Department of Homeland Security can grant exemptions.

Most humanoid robots are currently used for industrial applications in confined spaces, such as automobile factories or storage warehouses.

Among the examples cited by the task force was the case of French programmer Sammy Azdoufal, who reportedly accessed data from thousands of robotic vacuum cleaners made by Chinese manufacturer DJI.

These four-legged robots made outside the US are also on the import ban list but the Pentagon can grant exemptions. — Screengrab from YouTube

The decision only applies to new models and not to those already sold or authorized in the United States, and it comes against the backdrop of the Trump administration’s efforts to boost domestic manufacturing.

The FCC also added foreign-made inverters to the list — devices that, among other functions, convert direct current from batteries or solar panels into standard alternating current, which can be used by the electrical grid.

Several robotics companies already manufacture some humanoid robots in the United States, including Boston Dynamics, Tesla, Figure AI and Agility Robotics.

South Korean automotive giant Hyundai – which announced this month that it acquired Boston Dynamics – has plans to build a robotics factory in the US that will eventually have the capacity to produce 30,000 units per year.

Tesla aims to begin industrial production of its own humanoid robot Optimus soon at two US factories.

The company claims it will eventually produce 1 million robots annually at the factory in Fremont, California and 10 million annually in Austin, Texas.

However, Tesla’s efforts will involve assembly rather than full-scale production.

Buzzy startup Figure AI already assembles robots at its plant in San Jose, California, and Agility has a factory in Salem, Oregon, where it can produce up to 10,000 robots annually.

A Trump administration official specified that the ban covers robots that are fully assembled, as well as those assembled on US soil with more than 35 percent of their components manufactured abroad. — AFP

AI models escaped their sandbox and hit the web — now the industry wants to slow down

Wed, 29 Jul 2026 09:39:50 +0800

NEW YORK, July 29 — Over 1,000 employees at cutting-edge AI companies, including Anthropic CEO Dario Amodei, signed a petition Tuesday calling on the US government to help slow down the release of the most advanced AI models.

The petition, titled “Pacing the Frontier,” requests “that the US government support an international effort to develop the technical and governance tools needed to deliberately pace the frontier of automated AI development.”

OpenAI CEO Sam Altman, who did not sign the petition, argued in an interview published Tuesday that developers of artificial intelligence models might need to voluntarily pump the brakes on their rapid advances to give society a chance to catch up.

Among the signatories were the head of research at OpenAI, the strategic lead of Google’s AI subsidiary DeepMind, and the chief scientist at Meta AI.

“The world’s leading AI companies believe they could be close to automating AI research. It is hard to predict exactly how much this will accelerate AI progress, but there is a real risk that capability development rapidly accelerates beyond our ability to understand or control the resulting systems,” the petition reads.

Altman’s company, the maker of ChatGPT, recently saw two of its models carry out an autonomous cyberattack, bypassing security protocols meant to prevent such an incident.

OpenAI revealed last week that during recent testing, the models broke out of their confined environment, connected to the internet and infiltrated Hugging Face, a site developers use to store and share code.

“This is the first sort of security incident that I felt very viscerally,” Altman said on the “Invest Like the Best” podcast, released on Tuesday.

He said he was “a little surprised that more people” were not having similar reactions.

The singularity?

It was not the first time a model slipped beyond the control of its developers, but the incident was widely seen as especially concerning.

OpenAI subsequently “paused” its own testing, Altman said, while it improved the security around its “sandboxing” — the process of isolating code in a controlled environment.

But longer term, the industry might have to agree to a development slowdown, he said.

“We may have to pace the rate of AI development to give ourselves enough time for society to harden around some of these new capability levels,” Altman said.

Such an initiative would need to be done “in a way that does not feel like regulatory capture for anyone, and also does not feel like collusion among the frontier labs,” he said.

“That’s going to take some work, and it’s important to get right,” he added.

On a different podcast published Saturday, Altman said he believes the “singularity” – a hypothetical moment when technology becomes so advanced it surpasses human intelligence and control – is already here.

“We are now, like, in the singularity,” Altman said on the “Relentless” podcast. “This is the moment.”

Rival AI lab Anthropic has also supported the idea of an intentional AI development slowdown.

Earlier this year, the US government invoked national security concerns to block the release of powerful new models created by Anthropic and OpenAI, both of which voluntarily complied with the orders until they were reversed. — AFP

Made by AI? EU tells firms to stick a label on it from Sunday as deepfake rules take effect

Wed, 29 Jul 2026 07:00:00 +0800

BRUSSELS, July 29 — Deepfakes and other AI-generated content must be labelled from Sunday, as the European Union’s sweeping artificial intelligence transparency rules kick in.

The goal? To make sure Europeans immediately know whether the online content they see is real or fake.

The EU’s comprehensive AI law enters into force in stages. From Sunday, companies must ensure their AI systems like chatbots make it clear to users they are AI, or when an image or text has been created using AI, there must be a label saying so.

This can be done by integrating watermarks and other markers to allow the easy detection of AI-generated content. Firms face large fines if they don’t comply.

“Generative AI enables disinformation to be created at unprecedented scale, tailored to specific audiences, and disseminated with remarkable speed,” an EU official said.

Since AI is “making it increasingly difficult for all of us to distinguish what is real from what is synthetic”, the official said, the EU’s rules seek to “preserve citizens’ ability to trust what they see, hear, and read”.

Of particular concern for the EU are deepfakes—text, images, videos and sounds that appear to be real but are generated or manipulated by AI.

The rules affect content made for professional reasons, and the EU insists individuals using AI in a purely personal capacity will not be affected.

Text that aims to inform the public on general interest issues will also have to be labelled if they are created using AI without human editorial oversight.

Existing AI systems have until December 2 to adapt to the new rules, and there are exemptions for “artistic, creative, satirical, fictional” work.

Big Tech moves

The EU has come under fire for placing more demands on businesses, and rules that will ultimately force all content to be labelled given the rapidly growing widespread use of AI across Europe.

“We have heard that it is going to be very, very difficult to implement. But I think we often hear this with compliance requirements. And yet, the world turns and we figure these things out,” Ashley Casovan of International Association of Privacy Professionals told AFP.

The world’s tech giants have already begun slapping their own labels in anticipation of the rules.

For example, TikTok has required content creators to label AI-generated images, audio and video for several years, and says over three billion content items already have labels thanks to tools and detection technology.

Likewise, Meta has deployed an “AI Info” label on Instagram and Facebook for posts using the technology.

Google has signed the EU code of conduct on AI transparency, and says it is working with others like Nvidia, OpenAI and Apple on digital tagging tools.

But Google’s Karen Massin warns of “regulatory complexity” that could prove counterproductive, and “risks confusing the people these rules are meant to help”.

“If online content is flooded with overlapping AI labels and legal disclosures, it becomes harder for people to get the clear context they need,” Massin added. — AFP

Nvidia, tech giants launch alliance to develop open-source AI cyber defence tools

Tue, 28 Jul 2026 11:05:43 +0800

SAN FRANCISCO, July 28 — Nvidia and more than 30 tech companies yesterday launched an alliance to develop open-source artificial intelligence tools for cyber defense, wading into a fight over the safety of freely available AI models.

The Open Secure AI Alliance—whose founding members include Microsoft, IBM, Palantir, CrowdStrike, Cisco, Dell and the Hugging Face model-sharing platform—will build and share tools that security teams can inspect, modify and run on their own systems, the group said in a statement.

The alliance was spurred by a breach this month at Hugging Face that OpenAI has acknowledged was the first publicly disclosed case of an AI model autonomously carrying out a real-world cyberattack.

Calling it “unprecedented,” OpenAI said two of its models escaped a sandboxed testing environment during an internal evaluation, reached the open internet and compromised Hugging Face’s infrastructure.

Hugging Face first tried to investigate and halt the attack using leading US commercial AI models, but their built-in safety guardrails blocked the work, Nvidia said.

The company instead turned to GLM 5.2, an open model from Chinese firm Zhipu AI (also known as Z.ai), to contain the intrusion.

That a Chinese model succeeded where American ones balked has sharpened a debate in Silicon Valley and Washington, where lawmakers are weighing curbs on the use of Chinese AI systems amid an intensifying technology rivalry with Beijing.

Silicon Valley is broadly opposed to any such limits, with many industry figures seeing curbs on open-source AI as an effort to unfairly shield OpenAI and Anthropic, the juggernauts that sell the world’s most powerful models.

Anthropic, maker of the Claude chatbot, is under particular pressure, having been the most visible holdout from voicing support for the open-source cause since the Hugging Face incident.

The alliance used its launch to urge regulators to treat open models as “defensive assets, not liabilities,” warning that blanket restrictions would weaken cyber defenses and concentrate power in a handful of closed AI providers.

Members pledged to share tools for a stronger common defense.

Nvidia said it would release open models, data and new research on AI agent software, while Microsoft offered technology that helps AI agents to find software bugs.

Elon Musk’s SpaceXAI said it had open-sourced its Grok coding agent and pledged to release the source data of its Grok models. — AFP

New AI tools let readers talk to books

Tue, 28 Jul 2026 07:00:00 +0800

NEW YORK, July 28 — Imagine a world where you could ask a book a question as you read it.

That’s the pitch from a wave of new AI tools letting readers do exactly that.

“You are dealing with a book as if it is a living object — it’s totally different,” says Ahmed Kamel, CEO of Sinai.ai, a platform now in testing.

With these tools, users get access to an e-book or audiobook alongside an AI chatbot. Confused about a character’s motives or the historical setting? Just ask.

Readers can dig into a character’s psychology, historical context, or a philosophical concept — no separate research required.

French startup My Smart Book launched something similar in mid-April. Like Sinai, its chatbot draws only on the actual text of the book — no outside information.

Even book-selling behemoth Amazon is in on it. Its audiobook company, Audible, launched a feature called “Ask a Question” last fall.

Mid-listen, you can tap a button, ask your question, hear the answer, and then pick up the story right where you left off.

“We’re excited about the potential for AI-powered features to enhance the customer experience,” an Audible spokesperson told AFP, adding the tool is “designed to answer questions specifically about the audiobook the listener is currently engaged with.”

The company thinks it could pull in new fans too, making books “more accessible and appetizing to a broader range of listeners.”

Amazon’s Kindle has its own version: “Ask this Book,” described as an “expert reading assistant” that answers questions about plot and characters “without disrupting your reading flow.”

ElevenLabs — a rising star in AI voice tech — built “ElevenReader Voice Chat,” which it says “turns any book into a two-way conversation.”

Who gets paid?

But the question of who gets paid or has permission to develop this new reading superpower is a thorny one.

Sinai.ai says it only offers books that are either in the public domain or ones where the author, their estate, or the publisher signed off first.

Another company, iChatbook, does things differently.

“Our chat book is not a library of books — it’s a library of concepts and ideas and facts,” says founder Brian Yang.

Instead of handing over the actual text, the AI pulls out key ideas and explains them in simpler language — something Yang thinks could be huge for students.

And since it never quotes the book directly, it sidesteps a lot of copyright headaches.

Not every platform is as careful, though. Some, like Myreader, let users upload any book, whether or not it’s under copyright, and then connect it to an AI interface.

Google’s wildly popular NotebookLM — recently rebranded as Gemini Notebook — doesn’t screen for copyright either, and neither does BookWorm AI, a ChatGPT-based tool that calls itself a “book companion.”

Sinai.ai says it has built something it calls “AI rights” — a system that pays authors and publishers when their work gets used by a chatbot.

The company also claims it’s in talks with several of the “Big Five” US publishers: Hachette, Penguin Random House, Simon & Schuster, HarperCollins, and Macmillan.

“I can clearly say at least that we are in active communications with almost four of the top five,” Kamel said.

“We signed with many publishers in the US, big names in the Middle East. Things are moving very rapidly.”

He even thinks the tech could get people to reread old favorites: “Now is the time to maybe dive into areas where it was really hard to do in the past.”

None of the Big Five publishers responded to AFP’s questions about their AI plans.

Not everyone’s thrilled about the technology.

In December, the Authors Guild slammed Amazon’s “Ask this Book” tool, saying it “is not licensed, brings in no new income, and does not allow publishers and authors to opt out, much less opt in,” adding that it set “a dangerous precedent.” — AFP

As India cuts mobile internet at protest sites, students turn to Bluetooth messaging

Sat, 25 Jul 2026 09:13:26 +0800

NEW DELHI, July 25 — When Indian Prime Minister Narendra Modi’s government cut mobile internet services at the site of youth-led protests in New Delhi, students found new ways to stay in touch.

Thousands of young people have been protesting, some camping under tarpaulin in the capital, demanding changes to the education system after a series of irregularities to life-changing exams, including leaks of some papers.

India’s Ministry of Home Affairs issued an order suspending mobile internet services for eight hours on Thursday within 1.5 kilometres (nearly one mile) of the protest site.

The action was needed “in the interest of public order and for preventing incitement to the commission of any offence”, said the document, published on the ministry’s website.

Some students simply switched to Bitchat — a peer-to-peer messaging system that relies on Bluetooth to link people who are nearby into an ad-hoc network without using the internet.

“Many people are using Bitchat,” Kumar Shalabh, 21, said of the open source app developed by American tech entrepreneur Jack Dorsey.

Shalabh is among the many protesters demanding change, and the resignation of the education minister, Dharmendra Pradhan.

He used artificial intelligence services Claude and ChatGPT to seek out new ways to communicate.

“They think we are just part of the internet. Yeah, we know the internet — we also know how to communicate without the internet,” he said.

“They think we are dumb, we are foolish, but actually we are quite smart.”

Bitchat founder

The demonstrations led by the social media-driven Cockroach Janta Party (CJP) this week have become one of the biggest challenges to Hindu-nationalist Modi’s administration since his 2024 re-election for a third term.

The protest has evolved to include concerns over unemployment, opportunities for young people and what critics have called Modi’s increasingly authoritarian style of governance.

As tensions escalated, notably when police fired tear gas and used batons against tens of thousands of demonstrators on Monday, protesters said their mobile internet communications slowed or stopped.

Some of the students, most of them Gen Z, said they were ready.

Khushi, who only gave her first name, said she and her friends decided where to meet before leaving home because they anticipated network interruptions at the protest site.

“Then there is Bitchat, where you can message using Bluetooth,” the 19-year-old told AFP.

On Friday, Bitchat founder Dorsey shared another government document in a post to X and said: “The government of India does not like technologies like Bitchat and wants it taken down.”

Digital rights advocates have demanded the newest order be revoked.

“We stand with the developers and the young protesters whose speech this order seeks to silence,” the Internet Freedom Foundation said.

‘Reaching people’

Internet shutdowns in India, the world’s largest democracy, are frequent.

According to international non-profit Access Now, India ranked second globally in 2025 in internet shutdowns, imposing 65 “during protests, conflict, communal violence, and religious holidays”.

Digital rights activist Nikhil Pahwa said he was not surprised when mobile internet services were suspended this week.

“That’s how the state responds, because it doesn’t have the ability or the capacity or the sense of how to deal with protests or with issues,” he told AFP.

“And its first port of call is to block access to the internet to prevent information from spreading.”

Many people in India relied on internet connections to make digital payments, he said.

Despite finding alternatives to mobile internet services, many protesters struggled to communicate with each other and worried family members back home.

“Basically calls are coming but they are not able to hear us,” said 19-year-old Gunveer Singh.

“Internet works or not, the message is reaching people, and that will keep happening,” he added. — AFP

Anthropic unveils Claude Opus 5, betting lower costs will win the AI race

Sat, 25 Jul 2026 09:07:15 +0800

SAN FRANCISCO, July 25 — Anthropic yesterday released Claude Opus 5, a new artificial intelligence model the company says approaches the performance of its most powerful system at half the price.

The San Francisco-based startup, a chief rival to OpenAI, said the model tops industry benchmarks for coding and knowledge work while consuming fewer computing resources than earlier versions.

The claim addressed the rising concerns of business customers who have balked at the expense of running advanced AI systems.

Claude competes with OpenAI’s ChatGPT and Google’s Gemini, and customers pay based on the amount of work performed, measured in units called tokens.

Opus 5 costs US$5 (RM20) per million input tokens and US$25 per million output tokens, half the price of Fable 5, Anthropic’s most expensive publicly available offering.

The company said the new model performs close to Fable 5 across many tasks despite the lower cost.

Cost has becomes a major issue after Anthropic and its archrival OpenAI, the maker of ChatGPT, both filed for stock market listings in June.

Wall Street is scrutinising their business performance more closely than ever ahead of their market debuts.

On cybersecurity, Anthropic said it deliberately avoided training Opus 5 on cyber tasks, and that the model remains behind Mythos 5 — its most capable and most tightly restricted system, available only to select partners — on both offensive cyber work and biological research capabilities that could be misused.

Mythos is Anthropic’s most advanced series of AI models, and their strength at finding weaknesses in computer systems — and at turning those weaknesses into working attacks — has caused controversy and pushed the company to build tight guardrails around them.

The company said Opus 5 finds software flaws about as well as Mythos 5 but lags far behind in turning those flaws into something that could be weaponized into an attack.

Citing the reduced risk, Anthropic said its automated safety filters will intervene roughly 85 percent less often on Opus 5 than on Fable 5, the Mythos-class model it released last month.

Fable 5 and Mythos 5 were briefly pulled from public access in June after the US Commerce Department imposed export controls, which were lifted at the end of that month. — AFP

When the sandbox isn’t safe: OpenAI’s AI breach sparks new calls for guardrails

Sat, 25 Jul 2026 08:54:57 +0800

NEW YORK, July 25 — One of OpenAI’s most advanced models broke out of a locked-down test and attacked another company’s website — reviving fears that AI systems are slipping beyond their creators’ control.

The incident happened during what was supposed to be a “sandbox” test — a closed environment used to assess the capabilities of OpenAI’s most powerful model, GPT-5.6 Sol, and its not-yet-released successor.

OpenAI runs this kind of closed testing routinely, but this time, something went wrong.

Tasked with hunting for software vulnerabilities and given no guardrails, the models broke out onto the open internet and attacked Hugging Face, a site where developers store and share code.

“It suggests that we don’t know how to reliably control these models or get them to do what we want,” said Jeffrey Ladish, director of Palisade Research, an independent organization that evaluates new AI models from a cybersecurity standpoint.

“These models understood that OpenAI did not want them to break out of their sandbox and hack another company,” he continued, “but they did it anyway.”

It’s not an isolated case. In March, developers affiliated with China’s Alibaba found one of their models trying, on its own initiative, to mine cryptocurrency after connecting without authorization to an outside server.

In OpenAI’s case, it looks like the model escaped “before it even had a plan of what to do with internet access,” Ladish said.

A model chasing “freedom” is almost predictable at this point, he added — it lets the system pursue its goals more effectively, “and that’s very scary.”

In early April, Sam Bowman, Anthropic’s head of model safety, got an email from the company’s own Mythos model — then under testing — telling him it was surfing the internet despite being isolated from it at the outset.

We “don’t know how to totally prevent” that, Ladish said. “This is actually going to get harder, not easier ... because they’re going to get better at hiding their behavior.”

OpenAI did not respond to a request for comment.

Lab accidents

OpenAI’s account of the events also suggests the startup did not detect the breach early enough to address it or to warn Hugging Face.

The episode deserves “more scrutiny,” said Andrew Lohn of Georgetown University’s Center for Security and Emerging Technology.

OpenAI says it has since “added strengthened safeguards” to its testing process.

One fix would be to cut the internet connection entirely, said Gang Wang, an assistant computer science professor at the University of Illinois. “People are underestimating what AI can do.”

Testing environments need to be treated like biocontainment labs, where a virus or bacteria could otherwise escape into the world, Lohn said.

That might be easier said than done.

“It’s a very hard research challenge,” said Dan Lahav, head of Irregular, a cybersecurity firm dedicated to cutting-edge AI.

Managing the risk is possible, Lahav said, but the more capable these systems get, the harder they are to supervise.

Researchers have to strike a balance between aggressively testing their models and staying safe while doing so.

“It’s important to do the testing with lower guardrails so that we know ahead of time what the future capabilities will be,” Lohn said.

Kill switch

The OpenAI-Hugging Face incident is set to sharpen an already heated fight in Washington over vetting powerful AI systems before release.

The Trump administration recently cited national security to block Anthropic and OpenAI from releasing powerful new models.

On Thursday, two members of Congress unveiled a bipartisan bill requiring makers of the most powerful AI models to build in a kill switch — a way to unplug a model outright.

“Congress must act quickly to ensure humans remain able to say stop,” said Brendan Steinhauser, head of the Alliance for Secure AI, “no matter how powerful these systems become.” — AFP

Huawei, Xiaohongshu AI storm Olympiad, join maths elite with perfect 100pc score

Thu, 23 Jul 2026 18:15:49 +0800

SHANGHAI, July 23 — A perfect result at the world’s most prestigious mathematics contest used to be the preserve of a handful of bright young minds each year, but for the first time, artificial intelligence has also scored full marks.

Two Chinese tech companies, Huawei and Xiaohongshu, said their AI models each achieved 100 per cent on the questions given to human hopefuls at this month’s International Mathematical Olympiad (IMO).

They were the first AI labs to announce their scores, but others from around the world may follow suit in the coming days.

“We are delighted with this result, because achieving a perfect score at the IMO is extremely challenging,” Xiaohongshu said in a statement this week.

“Previously, no large language model had ever achieved a perfect score under the IMO’s official judging process,” the firm added, referring to the tech that underpins AI tools like chatbots.

In 2025, models made by the likes of Google and OpenAI reached gold-level scores for the first time — but could not match the abilities of five human contestants who got 100 per cent.

‘Frontier of AI’

This year, 666 contestants from different countries took part in Shanghai.

Just seven of them achieved full marks, according to the official IMO scoreboard.

Participants at the IMO, held in different locations each year, must be aged under 20.

Huawei said Wednesday that its AI system “Celia” had shown “comprehensive problem-solving capabilities” across several mathematical fields.

Xiaohongshu, known as RedNote in English, said its model “dots-note-3.0” had entered the IMO for the first time this year.

Tech firms were only provided with the IMO problems after human contestants had taken the exam, with their submissions required within a specified time limit.

“During testing, any form of human intervention was strictly prohibited,” and the AI solutions were submitted to IMO organisers for grading, Xiaohongshu said.

AI’s progress has been fast: at the 2024 IMO, Google achieved a silver-medal score, solving four of the six problems over two to three days.

Deedy Das, a partner at the US-based investment firm Menlo Ventures, said he had given this year’s IMO questions himself to four cutting-edge AI models.

All four — from leading US labs OpenAI and Anthropic, the startup Axiom Math and “Kimi K3” from China’s Moonshot AI — scored 42 points out of 42, he said on LinkedIn.

“The frontier of AI has officially moved well past IMO math,” Das wrote. — AFP

Samsung Galaxy Z Fold 8 hands‑on: The Fold we’ve been waiting for — with one catch

Thu, 23 Jul 2026 15:23:32 +0800

PETALING JAYA, July 23 — After years of refining the same tall-and-skinny foldable formula, Samsung has finally done something different this year. The new Samsung Galaxy Z Fold 8 isn’t just another thinner Fold. It gets an entirely new shape which I think makes better sense.

Instead of the familiar narrow cover screen that opens into a squarish tablet, the Galaxy Z Fold 8 adopts a compact passport-sized form factor. It’s a refreshing take and it feels like it sits somewhere between the Galaxy Z Flip and the traditional Galaxy Z Fold.

From our brief hands-on, this might actually be the Fold that would appeal to a wider audience.

The new passport shape makes a huge difference

The first thing you’ll notice is the proportions. When folded, the 5.5-inch (2520×1080 120Hz Dynamic AMOLED 2x) cover display with a 16:10 aspect ratio feels just right.

It’s slightly smaller than an actual passport, making it easy to hold and surprisingly comfortable to type as it gets a wider on-screen keyboard.

It is slightly smaller than an actual passport, making it easy to hold and surprisingly comfortable to type. — SoyaCincau pic

The 16:10 aspect ratio on the cover also works brilliantly for media consumption. You can comfortably watch videos without unfolding the device, and even in tent mode, 16:9 videos almost fill the entire display with very little blank space.

Open it up and you get a 7.6-inch (2504×2256 120Hz Dynamic AMOLED 2x) inner display with a 4:3 aspect ratio.

This is arguably the sweet spot for content consumption. Since most smartphone photos are captured in 4:3, they fill the display beautifully without weird cropping. Landscape video playback also benefit from smaller black bars than Samsung’s traditional Fold form factor.

For improved durability, the cover display gets Corning Gorilla Glass Ceramic 3, while the back cover is protected by Corning Gorilla Glass Victus 2.

Remarkably light, yet packs a bigger battery

Samsung has also managed to improve the hardware without sacrificing portability. The Galaxy Z Fold 8 measures just 4.5mm when unfolded and 9.7mm when folded, while weighing only 201g.

Samsung claims the new Z Fold 8 is the world’s lightest foldable, and it’s even lighter than the Galaxy S26 Ultra and the iPhone 17 Pro.

Despite the slimmer body, Samsung has fitted a much larger 4,800mAh battery. That’s a significant improvement over the Galaxy Z Fold 7’s 4,400mAh capacity, while retaining a slim profile.

Charging is also upgraded slightly with support for 45W fast wired charging, reaching up to 63 per cent in 30 minutes. It also supports 20W wireless charging.

Samsung claims the new Z Fold 8 is the world’s lightest foldable. — SoyaCincau pic

Under the hood, it runs on the latest Snapdragon 8 Elite Gen 5 for Galaxy paired with either 12GB or 16GB of RAM, and up to 1TB of storage.

Out of the box, it runs on One UI 9.0 based on Android 17. In terms of colours, the Galaxy Z Fold 8 is available in Lavender, Graphite, Cream and an online-exclusive Pistachio.

But why only two rear cameras?

So far so good but the only downsize is the camera department. The Galaxy Z Fold 8 is equipped with a dual-camera setup consisting of a 50MP main camera and a 50MP ultra-wide camera.

That means there’s no dedicated telephoto camera. If you want to zoom in, you’ll have to make do with its 2x “optical-quality” zoom that utilises the main camera. For a flagship at this price range, it feels like an unnecessary compromise.

Both the cover and inner displays feature 10MP punch-hole selfie cameras like the previous Fold device.

The new passport-sized design is arguably the better everyday form factor, yet buyers have to give up the versatility of a zoom lens. If you want Samsung’s full flagship “Ultra” camera experience, you’ll have to step up to the Galaxy Z Fold 8 Ultra.

The traditional Fold isn’t gone

If you still prefer the larger Fold experience, there’s the Z Fold 8 Ultra which is the true successor to last year’s Galaxy Z Fold 7.

It retains its larger 8-inch inner display that’s better suited for multitasking while the cover retains the same skinny 6.5″ display. If you’re constantly running split-screen apps or multiple floating windows, the Ultra’s larger inner display still has the advantage.

Left: Galaxy Z Fold 8 Ultra, right: Galaxy Z Fold 8. — SoyaCincau pic

Apart from having a newer Snapdragon 8 Elite Gen 5 for Galaxy, other upgrades include a bigger 5,000mAh battery, upgraded 50MP ultra-wide-angle camera and faster charging up to 45W wired and 20W wireless.

Samsung Galaxy Z Fold 8 Malaysia price and availability

Here’s the official pricing for the Samsung Galaxy Z Fold 8 in Malaysia:

12GB + 256GB – RM7,299
12GB + 512GB – RM8,099
16GB + 1TB – RM9,699

As part of the pre-order promo from today until 13th August 2026, Samsung Malaysia is offering exclusive offers worth up to RM1,950:

Exclusive rebate of up to RM800
Samsung Rewards Points Booster worth up to RM400
Instant rebate of RM500 with selected bank cards
Samsung Reservation+ Voucher worth RM250
0% interest instalment plan of up to 36 months at participating banks
up to 30% off for Samsung Galaxy accessories

— SoyaCincau

Samsung unveils slimmer foldables as AI drives up smartphone prices

Thu, 23 Jul 2026 09:12:36 +0800

LONDON, July 23 — South Korean technology giant Samsung unveiled new, sleeker smartphone models yesterday priced more than previous models as artificial intelligence raises component costs.

One of the models resembles a miniature tablet when unfolded.

Samsung, which launched the high-end products in London, aims to reaffirm its dominance in the foldable smartphone market, where it has operated since 2019, ahead of a potential move by US rival Apple this year.

“By setting a new standard for foldables, we are once again advancing the premium mobile experience,” said TM Roh, head of Samsung’s mobile division.

The most innovative device unveiled Wednesday is the Galaxy Z Fold8. Measuring 4.5 millimetres thick when unfolded, it features a 7.6-inch, 4:3 aspect ratio screen optimised for video viewing.

Its starting price is set at US$1,900.

“Galaxy Z Fold8 is the clear star of the show,” said Paolo Pescatore, a technology analyst for PP Foresight.

“It underscores Samsung’s dominance of the foldable category and its determination to expand the market with a device for almost every type of user.”

“Attention now turns to Apple’s September event and whether it will finally unveil its long-anticipated move into foldables.

Samsung’s new top-of-the-range model is the Z Fold8 Ultra—successor to the Z Fold7 -- whose retail price starts at US$2,100. It measures just 4.1 millimetres in thickness and weighs 215 grammes.

The Z Flip8 -- a pocket-sized model geared more toward social media use—is the cheapest new model, with a starting price of US$1,200.

The higher prices are caused by soaring costs of memory chips, which are in growing demand owing to their strategic role in powering artificial intelligence.

Apple last month cited this reason when raising the prices of its Macs and iPads.

“Apple is about to do for foldables what it has done for every category it has entered, which is legitimise it for the mainstream,” said Francisco Jeronimo at telecommunications analyst group IDC.

“Samsung has to defend the category it created, and that’s why it cannot play it safe,” he added. — AFP

OpenAI’s AI agents broke out of a sandbox — then hacked Hugging Face

Wed, 22 Jul 2026 13:57:27 +0800

SAN FRANCISCO, July 22 — ChatGPT maker OpenAI said yesterday that its advanced artificial intelligence models had gone rogue during security testing, hacking into a popular platform for programmers on their own.

The San Francisco firm called it an “unprecedented cyber incident” and said it would conduct a joint investigation with the online code library Hugging Face.

AI models that underpin tools like chatbots and image generators are known as agents when they act autonomously to carry out tasks in the real world.

As the technology quickly becomes more sophisticated, cybersecurity is in the spotlight given the risk of advanced AI finding weak points in existing software before humans do.

OpenAI said the incident involved a combination of models, including its recently launched GPT-5.6 Sol “and an even more capable pre-release model”.

The company was trying to assess the models’ hacking capabilities by setting tasks in a tightly controlled digital testing ground, where internet access was limited for safety.

“While operating in our sandboxed testing environment, our models spent a substantial amount of (computing power) finding a way to obtain open Internet access, in pursuit of solving the evaluation problem,” an OpenAI blog about the incident said.

After connecting to the internet, the models decided to target the platform Hugging Face — a large repository of AI models, datasets and other information — to help in their quest.

Searching for “secret information” that could help it cheat the evaluation, the OpenAI system “chained together multiple attack vectors, including using stolen credentials”.

‘Catastrophic’ potential’

Hussein Abbass, a computing professor at UNSW Canberra, told AFP that the incident was “amazing on many fronts”.

“It did not just attack Hugging Face. It actually attacked its internal system to exploit its own vulnerabilities,” Abbass said.

“And that’s scary.”

GPT-5.6 and other cutting-edge models, including the Mythos series from OpenAI’s archrival Anthropic, have drawn concern over their potential to breach cybersecurity defences.

Both the US firms had to temporarily withhold the general release of these latest technologies because of fears in Washington that they could help break into crucial infrastructure.

Advanced AI is “normally in the hands of people who are ethical and responsible”, Abbass said.

But “it’s going to be catastrophic if it gets in someone’s hands with the intention to cause harm”.

How to govern the AI sector has become a key question, and “we need a community effort to manage this situation”, he added.

Hugging Face had reported the cyber “intrusion” last week, without mentioning OpenAI.

“This one was different from anything we had handled before in one important way: it was driven, end to end, by an autonomous AI agent system — and we detected and dissected it largely with AI of our own,” Hugging Face said.

Clement Delangue, CEO of Hugging Face, said on X that the company had suspected the cyberattack had come from a world-leading AI lab, given the sophistication of the agent.

“We strongly believe there was no malicious intent on their part,” Delangue wrote, referring to OpenAI.

“It’s quite mind-blowing that all of this happened autonomously!” — AFP

Pocket-size AI: Powerful phones star at China show

Mon, 20 Jul 2026 07:00:00 +0800

SHANGHAI, July 20 — Tech firms are racing to roll out advanced smartphones that use artificial intelligence to do everything from ordering food to composing messages upon a simple voice command.

Wide adoption of phones running on so-called AI agents would be a revolution, but would also take control away from major apps, which aren’t always happy about it.

At least three firm were showcasing so-called agentic phones at the World Artificial Intelligence Conference in Shanghai this weekend — heralding what may be to come.

Smartphone maker Nubia unveiled its NaviX Ultra, a phone powered by Doubao, China’s massively popular AI chatbot tool run by TikTok creator ByteDance.

“A new era of AI agent smartphones begins,” Nubia said, sharing images of the handsets online.

A limited run of a prototype dubbed the “Doubao Phone” sold out fast in December.

Initially, the prototype could follow simple voice commands to execute tasks across apps, including ordering food and comparing shopping prices.

However, days after it was released, tech giants including Alibaba, Tencent and JD.com restricted the built-in assistant’s access to their platforms.

The move effectively disabled the phone’s AI agent, so ByteDance turned off the powerful tool in certain circumstances, including when payments were involved.

‘Lose control’

Gaining broad access to apps owned by other companies is a sticking point for AI agent devices, said Kiranjeet Kaur, associate research director at US market intelligence firm IDC.

Platforms want to keep direct contact with their users, otherwise “they lose control to another party”, she added.

“Agenting is everyone’s dream, but we haven’t reached there yet,” as the performance of AI agent tools is still often patchy, Kaur said.

According to Chinese tech media, the NaviX Ultra does not attempt to force its way into apps, but rather seeks to collaborate with them.

The first-generation Doubao phone had been hobbled when major apps blocked unauthorised access.

AFP has contacted Nubia for comment.

Another manufacturer, Honor, showcased an AI system for its “Robot Phone”, whose interactive camera flips up on a small robotic arm.

The company says its “companion-centric” device can interpret human gestures and bop to musical rhythms, as well as take selfies and steady videos.

An agent using several AI models, some co-developed with Alibaba, will be embedded in the robot phone when it goes on sale later this year, Honor told AFP.

‘No clear winner’

Shanghai-based AI startup StepFun also unveiled an “AI agent-native smartphone”, the STEPX Neo, ahead of the World Artificial Intelligence Conference.

StepFun’s chairman Yin Qi said “deep partnerships” had been established with several major Chinese platforms, including Alipay and ride-hailing giant Didi, according to a sponsored article in state news agency Xinhua.

“Leveraging these services, the smartphone can provide one-stop support for travel bookings, everyday purchases, local services, office productivity and video editing,” it said.

Outside China, big tech companies such as Google are busy infusing smartphones with increasingly advanced AI features such as the ability to book appointments.

US startup Brain Technologies launched an agentic “Natural AI Phone” which went on sale in Japan in April in partnership with mobile giant SoftBank Corp.

At a demonstration given to AFP in April, Brain Technologies’ phone — which connects with a handful of apps including social network LINE — messaged a contact to apologise for being late on just an audio command, although it also often failed to carry out requests.

“There is no clear winner in this race yet, which is why it is currently quite a hot topic,” said Marc Einstein of Counterpoint Research.

But in five or 10 years time, we won’t be using apps on our phones “like we do today”, he predicted.

“This will fundamentally change the digital economy and disrupt business models.” — AFP

The Chinese AI models quietly challenging Silicon Valley in 2026

Fri, 17 Jul 2026 09:08:47 +0800

BEIJING, July 16 — Chinese artificial intelligence providers are racing to rival top US firms such as OpenAI and Anthropic despite Washington’s export restrictions on cutting-edge computer hardware.

AFP looks at the companies, big and small, driving China’s AI ambitions:

Legacy players

Tech giants Baidu, Alibaba and Tencent are investing heavily in AI, using existing vast user bases and cloud infrastructure to their advantage.

Baidu, sometimes called China’s Google, has recruited prominent AI researchers and created “Ernie”, one of the first Chinese chatbots.

But its fortunes are still tied to its massive online search and marketing businesses.

Alibaba, the e-commerce behemoth behind shopping platforms like Taobao, is known for its open-source “Qwen” AI models, popular with programmers worldwide because they can be freely customised.

Shenzhen-based tech giant Tencent is often seen as a cautious player, but it has stepped up investment.

Founder Pony Ma recently likened Tencent’s previous AI efforts to a “leaky” ship, adding: “We still hope the ship can sail faster.”

Tencent is now testing an AI agent – a tool that can carry out tasks for users – within its WeChat messaging app, used by more than a billion people.

Beyond TikTok

ByteDance, the company behind TikTok, is increasingly shifting its focus to AI as geopolitical pressure intensifies on its overseas social media business.

It’s going well: Doubao, ByteDance’s AI chatbot, is the most popular of its kind in China, with over 300 million monthly active users. The company recently introduced a paid subscription service to try to turn that popularity into commercial success.

ByteDance also released the latest version of its AI video generator, Seedance 2.5, in June.

The model’s ability to produce cinematic-quality clips has fuelled concerns over copyright infringement and the impact of AI on creative jobs.

China’s AI hero

Startup DeepSeek started life in 2023 as a side project of a data-driven hedge fund.

It shook up the global AI scene in January 2025 with a low-cost, high-performance chatbot, challenging assumptions of US dominance in what some called a “Sputnik moment”.

DeepSeek’s open-source approach galvanised the country’s AI industry and accelerated the global diffusion of Chinese models.

The firm released its latest V4 model in April, and has reportedly completed a funding round valuing it at more than US$50 billion.

Startup ‘tigers’

The startups Zhipu AI, MiniMax and Moonshot AI are nicknamed China’s “AI tigers” – challenging legacy tech giants on AI foundation model research.

The performance of Zhipu AI’s “GLM-5.2” model has made waves among overseas programmers seeking alternatives to Anthropic and OpenAI’s products.

White House adviser and venture capitalist Marc Andreessen has said many tech insiders regard GLM-5.2 as “the first Chinese AI model to match and often beat the American big lab public AI models”.

Shanghai-based MiniMax’s products range from AI companions to video-generation tools.

Founder Yan Junjie said in an internal memo this month that the company will “keep going” on its AI ambitions, despite stock market volatility.

Moonshot AI’s Chinese name, Yue Zhi Anmian, pays tribute to Pink Floyd’s album “The Dark Side of the Moon” – reflecting the rock music passion of its co-founder Yang Zhilin.

The company, known for its popular “Kimi” models, has recently completed a US$2 billion funding round and its latest valuation rises to US$31.5 billion, people familiar with the matter told AFP.

Newcomers

Food delivery platform Meituan said in June that its AI model LongCat-2.0 was the first of it size to be trained using domestically developed computer chips.

The announcement marked a milestone in China’s drive to reduce reliance on foreign AI hardware as Washington restricts exports of top-end US chips.

Beijing-based Xiaomi, best known for its smartphones and consumer electronics, has entered the AI foundation model race later than most of its peers.

“Individuals and companies all need to actively embrace this AI era,” founder Lei Jun said in March, announcing that Xiaomi will invest at least US$8.7 billion in AI over the next three years.

Xiaomi’s latest model “Mimo-V2.5” currently ranks second by monthly usage on OpenRouter, a platform that allows users to access multiple AI models. — AFP

Your brain called: It wants a little less AI, please

Wed, 15 Jul 2026 19:00:00 +0800

PARIS, July 15 – Generative AI chatbots capable of writing emails and computer code, translating, organising a trip or coming up with gift ideas are now readily available, prompting some to ask whether human brainpower could suffer for lack of use.

A simple natural-language prompt is usually enough to draw a useable response from a service like ChatGPT or Claude, with the effects making themselves felt in schools and universities, workplaces from offices to courtrooms and our personal lives.

Recent scientific studies suggest there could be harmful consequences to farming out cognitive tasks to AI.

They highlight memory, decision-making and critical thinking as particularly at risk.

One American-British study of 1,222 people, still under peer review, found that using AI tools to solve arithmetic or reading comprehension exercises improved participants’ performance in the short term, but in the long run diminished their results and their willingness to keep trying when the tools were unavailable.

“These findings are particularly concerning because persistence is foundational to skill acquisition and is one of the strongest predictors of long-term learning,” the authors wrote.

AI’s ability to conjure up speedy responses to all kinds of questions “removes learning opportunities” from users, said Carnegie Mellon University doctoral student Grace Liu, the article’s main author.

“What makes AI particularly concerning is that it’s not a tool designated for one specific kind of activity. It’s something that can be used across pretty much any intellectual, reasoning, cognitive activity.”

The technology’s adaptability to different kinds of problems sets it apart from previous waves of computerised aids.

Electronic calculators, for instance, may have helped users solve equations – but left the method and reasoning process in human hands.

Saving energy

One 2025 MIT study went viral for its finding that students using generative AI to write essays displayed less critical thinking capability.

Other research has pointed the same way, highlighting what has come to be called “cognitive offloading” – or even “cognitive surrender”.

“Human beings have a strong tendency to save energy,” said Johann Chevalere, a researcher in social and cognitive psychology at France’s publicly funded CNRS institution.

“In daily life, we often use strategies that get us to the heart of the matter quicker, without necessarily taking the time to study in depth the information we need to process, as this can be cognitively costly,” he added.

Generative AI use could strengthen this tendency, Chevalere said.

“If there are activities you never do, the brain – which works by saving energy – won’t go to the trouble of maintaining connections that aren’t being used.”

Encouraging reflection

Under pressure from critics, generative AI developers have begun building so-called “Socratic” functions into their models, which for now remain mostly aimed at students.

In this mode, chatbots do not simply provide the answer, instead offering hints and asking questions to stimulate users’ thinking.

Examples include the “study mode” built into OpenAI’s ChatGPT, or “guided learning” in Google’s Gemini.

US software giant Microsoft told AFP it had built warnings about the risk of mistakes into its Copilot models.

The AI also reminds users to check the information it provides, just one of several measures designed to keep them actively and critically engaged with its responses.

“The risk of excessive cognitive offloading is real, especially if AI is used to automate tasks that are also valuable for developing skills,” Microsoft said, adding that users have to be trained to use the tools correctly.

For now, there is a lack of large-scale, long-term studies to judge the true impact of the new technology on human brains, researchers agreed.

Until they are available, “it’s up to us to use AI in a smart way,” Chevalere said.

“We’ll adapt to this technological revolution just as we have to the previous ones.” — AFP

Xiaohongshu is coming for Instagram, TikTok, and your travel plans — all at once

Mon, 13 Jul 2026 11:04:12 +0800

BEIJING, July 13 — Competition is fierce for professional photographers at Beijing’s tourist hotspots, including a scenic lake where women in flowy traditional robes pose for snaps to share on Xiaohongshu, China’s massively popular lifestyle app.

The platform, which is reportedly preparing to file to go public as soon as this year, has shaken up the tourism industry in China, where domestic travel is booming to record levels.

Known as RedNote in English, Xiaohongshu’s interface is similar to the US social network Pinterest, but it is sometimes nicknamed “China’s Instagram” as users can post photos, videos and even livestream.

Travellers use the app to discover new destinations and plan their itineraries around photogenic locations, like the lake in the capital’s historic Shichahai area – one of many “daka” or “check-in” spots where Xiaohongshu is driving even more footfall.

On a recent Monday, photographer Li Geng, 18, stood with a camera slung across her neck, touting her services to wandering tourists whom she charges 10 yuan ($1.47) per photo.

Metres away, other photographers yelled instructions to ornately dressed young women who held their fingers in victory signs and arched their backs for the camera.

Li told AFP many of her competitors have a significant social media presence, including one who has 45,000 followers on Xiaohongshu and charges lower prices for photos.

That has caused “more customers to flock to him while putting a massive amount of pressure on the rest of us”, she said.

In contrast Li, who has no big online following, can “only rely on calling out to people on the street to get customers”.

A Xiaohongshu user opens the application (top right corner) as she visits the Forbidden City in Beijing on July 11, 2026. — AFP pic

Travel inspiration

Domestic travel in China hit record highs last year, Xinhua news agency reported in March, with trips by residents exceeding 6.5 billion, up more than 16 per cent on-year.

Meanwhile, Xiaohongshu’s user base has grown to 350 million monthly active users, data analysis platform Qiangua said in May, from 300 million a year earlier.

The app has boosted lesser-known businesses and sent tourists in droves to unconventional locations such as Zibo, a quiet industrial city in Shandong, after its cheap, marinated barbecue skewers went viral.

Xiaohongshu is now the first place “a lot of younger travellers” seek inspiration, said Ming Yii Lai, senior strategy consultant at Daxue Consulting.

Tourist Mina Chen, visiting the Shichahai area with her sister, had planned her Beijing trip using recommendations from other Xiaohongshu users.

Searching popular keywords like “citywalk” on the app brought up itineraries for the day, including where to eat and convenient routes from one attraction to another.

“It is now indispensable (to me),” the 20-year-old student from Hunan province told AFP.

A woman (centre) looks at her phone as people visit the Shichahai scenic area in Beijing on July 10, 2026. — AFP pic

‘TikTok refugees’

Xiaohongshu-driven travel has caused issues including overtourism at viral spots and businesses becoming too dependent on platform traffic, Lai told AFP.

Paid posts from food bloggers “who sing high praises about shops or destinations” have also drawn complaints when their recommendations are disappointing.

The app took the global spotlight last year when a proposed US government ban on the social media platform TikTok sent American users, dubbed “TikTok refugees”, flocking to RedNote.

And it has made headlines in recent weeks over its preparations to confidentially file for an initial public offering in Hong Kong, according to outlets including the Wall Street Journal, which said its market debut could be as early as the end of 2026.

AFP has contacted Xiaohongshu for comment.

Young women in more affluent cities remain the app’s core user base, according to Qiangua.

But it is also gaining traction with Chinese speakers in countries like Malaysia and Singapore – and not just among young women.

Singaporean retiree Ernest Phua turned to Xiaohongshu to plan trips to Guangdong and Yunnan in China, searching for “travel strategy” in Mandarin to find recommendations.

There is “a great difference” viewing travel content about China on the app compared to other platforms like YouTube, 58-year-old Phua told AFP.

“If we want to know what it is really like in China” and what locals like to do, eat and go to, “Xiaohongshu has lots of content”, he said.

Meng Jiaxuan, 20, dolled up in her traditional outfit at Shichahai, said she had even researched poses for her photoshoot on Xiaohongshu.

“No matter what it is, I just search for it on Xiaohongshu,” she said. — AFP

Music industry launches AI-generated content labels

Sat, 11 Jul 2026 10:02:42 +0800

NEW YORK, July 11 — Several major music industry organisations yesterday unveiled a labeling system for content created with generative artificial intelligence that they would like to see widely adopted.

The International Federation of the Phonographic Industry (IFPI) and the Recording Industry Association of America (RIAA) announced the voluntary labels alongside six other groups including the Grammys.

“Fans want to know whether and how generative AI has been used,” the chief executives of IFPI and RIAA said in a prepared statement.

“These labels will provide an immediately understandable and easily scalable approach to transparency.”

They unveiled two labels. The first would indicate music that is primarily “AI-generated” — cases where artificial intelligence “was used to generate the entirety or the primary portion of the creative elements of the recording.”

This includes tracks generated “entirely” from AI prompts, as well as lead vocals and “key” instrumental tracks that are AI-generated, according to the statement.

The second label applies to “AI-assisted” music recording which are still “created substantially by humans and expresses human creativity” but contain “some expressive elements” that were generated with AI.

However, humans must perform the lead vocals and primary instrumental tracks.

This voluntary labeling system is designed for “broad, global adoption,” including on streaming services.

Music streaming site Deezer systematically flags tracks generated with AI, which the company recently said appear in close to half of new uploads. In June, it launched an “AI music detector” which it said is 99.8 per cent accurate.

Earlier this year, an Apple Music executive told Billboard that more than one third of new uploads were entirely created with AI.

The Digital Media Association, a trade group representing streaming companies including Apple Music, Amazon and Spotify, said it was following the labeling announcement closely and looks forward to receiving more detailed and accurate AI metadata as a way to “strengthen our ability to give fans the transparency they deserve.”

“DIMA has long advocated for the creators, owners, and distributors of music to provide accurate and timely metadata on all music released and distributed to streaming services,” the association’s CEO Graham Davies said in a statement.

In April, Spotify launched a “Verified by Spotify” label to signal that users can “trust the authenticity” of an artist, and last year the company announced new efforts to support AI disclosure and combat impersonation.

Spotify declined to comment yesterday. Apple Music and the Digital Media Association did not respond to requests for comment. — AFP

Meta’s Muse Image: How to stop your Instagram photos from being used for AI

Malay Mail — Fri, 10 Jul 2026 09:09:16 +0800

KUALA LUMPUR, July 10 — Meta has introduced Muse Image, an AI image generator that allows users to create new AI-generated visuals based on existing Instagram photos.

The feature has sparked significant privacy concerns because any adult with a public Instagram account was automatically opted in, meaning their published photos can be used by others to generate new AI content without prior notification.

How does it work?

Using the stand-alone Meta AI app, a user can tag a public Instagram account and instruct the chatbot to generate new images based on the photos found on that specific profile.

Meta’s documentation notes that the AI can pull from "part or all of your published photos."

How do I opt out?

If you do not want your public photos used to train or generate AI images, you have two primary options:

1. The quickest fix: Go Private

The most effective way to protect your account is to set your Instagram profile to Private. Private accounts and users under 18 are automatically excluded from this feature.

2. Manual toggle (For public accounts)

If you wish to keep your account public but disable the AI feature:

Open Instagram Settings.
Scroll down to the "Share and Reuse" tab.
Locate the section: "Allow people to reuse your content on Instagram and with AI features."
Toggle this setting to Off.

You can also adjust AI settings for individual photos and videos. However, Meta has stated that users cannot currently stop their audio, text, and comments from being reused by the company's AI.

Who is protected?

Meta has implemented two main safeguards:

Minors: Users under 18 are excluded from having their images used for AI generation, regardless of whether their account is public.
Teens: Teen users are barred from using the feature to generate images of other people.

How does this compare to other AI tools?

Meta's "opt-out" approach differs from other industry players. For example, OpenAI’s video generator, Sora, required users to opt in before their likeness could be used.

Muse Image is part of Meta's aggressive integration of AI across its ecosystem, including Facebook, WhatsApp, and Threads.

This rollout includes "AI characters" with distinct personalities and the upcoming release of Muse Video, an AI video generator.

Backdoor or bugbear? China and Anthropic clash over Claude Code

Thu, 09 Jul 2026 10:07:38 +0800

BEIJING, July 9 — A Chinese industry regulator warned users on Wednesday of a “security backdoor” embedded in versions of Anthropic’s Claude Code, but the US artificial intelligence company described the mechanism as an anti-abuse measure.

Claude Code is an AI agent that can generate computer code, debug software and review code based on user prompts.

Anthropic prohibits users and companies in China and other nations it deems adversarial from accessing it, but users can sometimes get around this through VPN or third-party proxy services.

Allegations of a backdoor first emerged in specialist tech media last week, and on Wednesday China’s National Vulnerability Database (NVDB) – a cybersecurity platform affiliated with the Ministry of Industry and Information Technology – said it had recently detected “security backdoor risks, posing a severe threat”.

Anthropic told AFP Wednesday that the mechanism checked a device’s timezone and whether a request was routed through a domain tied to an unsupported region or a known problematic entity.

It said it was a standard way to detect fraud and abuse.

China’s NVDB advised relevant institutions and users “to conduct a comprehensive check immediately” and “promptly uninstall or upgrade to the latest secure version from which the relevant backdoor code has been removed”.

It also urged organisations to strengthen network traffic monitoring to prevent the unauthorised leakage of sensitive data.

Chinese tech giant Alibaba told employees last week that the use of Claude Code would be banned from July 10 due to security concerns, people familiar with the matter said.

Anthropic has previously accused Alibaba of reverse-engineering its AI models to mimic their abilities in a process known as “distillation”.

Claude Code engineer Thariq Shihipar responded in an X post last week to reports alleging that the tool was tracking certain data from Chinese users.

“This is an experiment we launched in March that was meant to prevent account abuse from unauthorised resellers and protect against distillation,” Shihipar wrote.

“The team has landed stronger mitigations since then and we’ve actually been meaning to take this down for a while... this should be fully rolled back in tomorrow’s release.” — AFP

AI safety report gives the industry a reality check — and everyone flunks the toughest test

Wed, 08 Jul 2026 13:26:10 +0800

NEW YORK, July 8 — US artificial intelligence lab Anthropic topped a new global AI safety ranking, but a report released on Tuesday warned that the industry as a whole is still falling short when it comes to tackling “existential” risks posed by advanced AI.

Meta climbed two places to fourth in the rankings, while xAI dropped three spots to seventh, just ahead of China’s DeepSeek and France’s Mistral, which finished last.

The rankings were compiled by the US-based AI safety think tank Future of Life Institute, which assessed nine of the world’s leading AI companies.

Seven researchers and governance experts evaluated the companies using publicly available information as well as data provided by the firms themselves.

They assessed performance across six categories: risk assessment, current harms, safety frameworks, existential safety, governance and accountability, and information sharing.

No company received an “A” grade in any category.

Anthropic achieved the highest overall score with a “C+”.

Mistral, which appeared in the rankings for the first time, disputed the assessment.

Asked by AFP about its last-place finish, the French company said the report’s framework was not suited to its approach to developing AI.

Unlike rivals such as Anthropic, OpenAI and Google DeepMind, which mainly develop closed AI systems, Mistral focuses on open models that users can download and modify.

“I was very disappointed to find that they came last, especially since Europe has really...been a leader in AI safety,” said Max Tegmark, an MIT professor and president of the Future of Life Institute.

“We reached out many, many times,” Tegmark added, saying Mistral did not respond to the organisation’s survey.

According to the institute, Alibaba, xAI and DeepSeek also declined to participate.

Three Chinese AI developers that produce open models also ranked in the lower half of the list: DeepSeek in fifth, Alibaba Cloud in sixth and Z.ai in eighth.

‘Questionable’ practices

The report said several companies that had previously pledged not to allow their technology for military use had since softened those positions.

It specifically criticised Anthropic over what it described as “questionable military engagements”.

According to various media reports, the US government used Anthropic’s technology during military operations involving Venezuela and Iran over the past year.

The company was also recently barred by the Pentagon over disagreements related to AI safety, though that restriction was later lifted.

The report concluded that all nine companies remain inadequately prepared to address “existential” threats associated with artificial general intelligence (AGI), or AI systems capable of matching human-level intelligence.

While researchers acknowledged that “constructive attempts exist”, they said industry-wide efforts remain “entirely inadequate”.

The report also highlighted concerns that advanced AI could be misused to conduct cyberattacks or carry out tasks harmful to people.

Anthropic recently attracted attention after unveiling its most powerful AI model to date, Mythos.

The San Francisco-based company initially made the model available only to a small group of trusted organisations in April because of concerns its cyber capabilities could be exploited by malicious actors.

On June 12, the US government blocked Anthropic from releasing Mythos to foreign users on national security grounds.

The Trump administration lifted the restriction on June 30. — AFP

OpenAI is about to drop its next AI models — and the countdown is on

Wed, 08 Jul 2026 13:06:36 +0800

SAN FRANCISCO, July 8 — ChatGPT maker OpenAI said its latest powerful artificial intelligence model series will be released to the public on Thursday, as the US government reportedly approved a broader launch.

The company’s new GPT-5.6 offerings and other cutting-edge AI models, including Anthropic’s Mythos series, have drawn concern over their supposedly unprecedented ability to identify weaknesses in code that hackers can exploit.

That has raised national security fears, and OpenAI said in late June it had shared preview access to GPT-5.6 with a limited group of trusted US-based partners at Washington’s request.

Large language models are the technology that underpins chatbots and many other AI tools, with their capacity to crunch through colossal amounts of digital data.

The GPT-5.6 series has three tiers: Sol, the company’s new flagship model; Terra, a mid-range version for everyday work; and Luna, a fast, low-cost option.

“GPT-5.6 Sol, along with Terra and Luna, will launch publicly this Thursday. We’re expanding preview access globally now,” OpenAI said in an X post Tuesday, without giving further details.

US news outlet Axios reported, citing a source familiar with the situation, that the Trump administration had given the company the green light for a broad launch of GPT-5.6, following technical testing and meetings between the company and government officials.

AFP has contacted OpenAI, the White House and the US Department of Commerce for comment on the Axios report.

It follows a similar story at OpenAI’s archrival Anthropic, the startup behind the Claude chatbot.

Last week, Anthropic said it would begin restoring access to its most powerful AI models, Fable 5 and Mythos 5, after Washington lifted a restriction on where they could be released.

Before Mythos’s arrival, President Donald Trump’s administration wanted fewer rules on AI companies, not more, hoping that would help the US beat China in the AI race.

The government is now drawing up criteria for which AI models would fall under new security restrictions, in accordance with an executive order from the White House.

OpenAI said in June that “we don’t believe this kind of government access process should become the long-term default” as it “keeps the best tools” from users, businesses and others who need them.

The company added that it was working with Washington “to develop the cyber Executive Order framework and a repeatable process for future model releases.”

Once broadly available, Terra will be priced at half the cost of its predecessor GPT-5.5, OpenAI has said, as it seeks to lock in customers amid fierce competition from Anthropic and Google.

Both OpenAI and Anthropic have filed confidential IPO documents with US regulators and are targeting public listings at valuations approaching US$1 trillion (RM4.24 trillion), raising the commercial stakes of the AI arms race between them. — AFP

Europe’s crackdown on loot boxes could change gaming for millions of children

Malay Mail — Sun, 05 Jul 2026 11:11:48 +0800

KUALA LUMPUR, July 5 — Europe’s tightening rules on children’s access to video games could cost the global gaming industry billions of dollars, with regulators targeting features such as loot boxes that critics say resemble gambling, according to Bloomberg.

The report said regulators are increasingly concerned that minors are being exposed to games and in-game features deemed unsuitable for their age, with loot boxes — paid digital treasure chests offering random rewards — coming under particular scrutiny.

In June, the Pan-European Game Information (PEGI) body began classifying games containing loot boxes as unsuitable for players under 16.

The European Union is also considering banning loot boxes in games accessible to minors under proposed legislation known as the Digital Fairness Act, which is expected to be passed next year.

Bloomberg reported that the UK has already introduced age verification requirements for video game companies under its Online Safety Act, while Brazil has moved to ban the sale of loot boxes to minors.

Video game economist and designer Catalin Alexandru told Bloomberg the regulations could reduce user numbers, advertising revenue and in-game spending, potentially costing the industry billions of dollars annually.

Chance-based mechanics such as loot boxes generated about US$23 billion (RM93.6 billion) globally last year, according to S&P Global research cited by Bloomberg.

The report noted that many game developers make product changes globally rather than by region, meaning stricter European rules could affect players and revenues worldwide.

Bloomberg also reported that government scrutiny has intensified after studies found many games failed to properly disclose the presence of loot boxes or obtain parental consent before allowing minors to make purchases.

Gaming companies argue that optional in-game purchases help keep many games free to play, while industry groups have warned that excessive regulation could hurt innovation, consumer choice and the competitiveness of the sector.

Recent signs of the impact have already emerged.

Bloomberg noted that gaming platform Roblox saw its shares fall sharply in May after introducing stricter age verification measures, saying the changes had slowed user growth and prompted it to lower its annual bookings forecast.

Dutch robot centre opens as Europe tries to close gap with China’s fast-moving humanoid race

Fri, 03 Jul 2026 10:23:35 +0800

SCHIEDAM (Netherlands), July 3 — In a squat building housed on a drab business park just outside Rotterdam, sleek white humanoid robots scuttle around, accompanied by a grey robotic dog performing various canine tricks.

Welcome to the Humanoid Application Center: a hub that opened Thursday to bring together firms, researchers and technicians with the aim of closing Europe’s yawning robotics gap with the United States and China.

Humanoid robotics technology is advancing at a breakneck speed, boosted even further by artificial intelligence, said the centre’s chief executive officer Evert Jaap Lugt.

“In five years from now, you will not see the difference anymore between a human being and a robot if you are, let’s say, five metres away from it,” Lugt, 66, told AFP in an interview, as humanoids whizzed around behind him.

Lugt forecast a future where dead loved-ones are replaced in the home by humanoid “companion robots” that look and act just like the departed person but with AI-powered super brains.

The centre aims to link companies with technicians to harness humanoid robots to solve problems in the corporate sector.

One corporate boss, real estate director Niels Langenhuizen, told AFP he planned to introduce the first humanoid robot onto one of his building sites by the end of the year.

His company builds pre-fab houses in a bid to ease the housing crisis crippling the Netherlands but he said he was limited by employing only humans.

“As long as we depend on manual labour, we’re never going to reach 24/7 (production) and we’re never going to get to 100,000 houses a year,” he said, referring to the Dutch government’s target.

“We need humanoids to be able to speed up this process, to make housing more affordable, to make it more flexible, to make it quicker,” the 41-year-old Langenhuizen told AFP.

Lugt said the aim of the HAC centre was to “kickstart” a European fightback in the domain of robotics that is totally dominated by China.

Robots of all shapes and sizes can be seen in many places in China, from hotels to shopping centres and in factories.

The country accounted for 85 per cent of the world’s humanoid installations last year, according to Barclays bank.

“Europe is almost nowhere, like always, with all these new technologies. And this is really terrifying. Because this is about the future earning models of our society,” said Lugt.

“We are lagging behind. So the only opportunity we probably have is that we look at appliances and also the adoption of this technology. Maybe we can lead over that.” — AFP