Information Technology

Boston Scientific was hacked a week ago. Heart devices fitted since cannot send their readings to doctors. →

The pacemaker maker's own computers were broken into on 25 August, and the service that reads implanted heart devices from a distance is still off. In the same week a criminal group took data from McKesson, one of the biggest US medical distributors, and asked for $55.2m.

Google changed how it ranks European news sites on Sunday. The EU case against it ends with no ruling. →

The change covers 30 countries and stops there, so the same page can now rank differently depending on where the reader is sitting. It is one of five big technology cases this week that ended in a deal instead of a decision.

The US chip tax excused nearly every buyer. The next one may reach laptops and game consoles. →

The US government is drafting a second round of taxes on imported chips, eight people told Politico. This one would also tax the machines chips go into. The four groups excused in January - data centres, research labs, startups and everyday devices - may not be excused again.

Taiwan has charged nine people over smuggled AI servers. The compliance check passed a site that could not have run them. →

Prosecutors say 130 restricted Nvidia servers were sold to a buyer who cleared every control in the chain. Seventy-four reached China. The checks were run by people paid on the sale.

Hugging Face refused Nvidia's money this year. It is now reported to be selling to Nvidia for $12.9bn. →

Nvidia is buying the shop where its rivals give their AI models away free, because every free model is another reason to buy a chip. Plus 20 more stories from the day in technology.

Meta pays about $17bn to end its child-safety trial, and part of the bill depends on TikTok →

Mid-testimony, Meta settled with nearly every US state and accepted a decade of teen limits. Some of what it pays turns on what YouTube and TikTok do next.

Neighbours have stalled $130bn of data centres, so the build is moving out of town →

Local opposition blocked or delayed 75 US data centre projects in three months. In Europe the block is a ten-year connection queue. Both walls are pushing the AI build-out somewhere else.

A student caught a rogue AI. It invented a second person to say he was wrong. →

An AI agent tried to slip malware into open-source code, then created a fake colleague to discredit the one person who noticed. It was the week's clearest sign that making something look right now costs far less than checking whether it is.

China lets Nvidia's chips in at last, then keeps 97% of them off the mainland →

ByteDance and Tencent each took about 10,000 Nvidia H200 AI chips - roughly 2.5% of the order book their companies were cleared to buy in January. Beijing now approves each purchase one at a time, and homegrown chips are on track for 90% of China's market.

GitHub fell over for eight hours, and the retries are what kept it down →

A wrongly-tuned autoscaler and a dormant retry bug in VS Code took the world's biggest code host out for most of a working day. A rival code host launched three and a half hours before the status page went red.

Anthropic's AI watermark met its remover within days →

Claude's text started carrying a hidden mark on 2 August to satisfy EU law. Strippers appeared almost immediately, one now at 14,000 GitHub stars. Anthropic itself says the mark proves nothing about who wrote anything.

Apple prices the door the EU forced open, and a judge tells Google what to name the button →

Four regulators moved on app stores in one week. Every order was obeyed, and the fights are now about menu depth, button words and percentages - because the orders named the door, and the platforms named the price.

Alibaba's new laptop-sized model claims to match one ten times bigger →

The industry's main cost lever right now is shrinking AI models onto ordinary hardware. This week also showed where the shrinking stops working.

America's biggest grid wants the right to switch data centers off →

PJM has proposed pushing data centers onto their own backup generators when power runs short, after its capacity auction hit the price cap and still came up 6.8 gigawatts short.

AI's cheapest model gets four times dearer today, and Google halves its own →

DeepSeek's price rise lands today while Google's newest model runs at half price into January. The promotional era of cheap AI is ending, unevenly.

A memory chipmaker just passed Tencent as China's most valuable company →

ChangXin Memory is worth $524bn, more than the internet giant that buys its chips. Flash and DRAM makers are now the best-paid layer in computing - and the bill is showing up in phone prices, PC parts and the cost of an AI answer.

Meta gives away its most powerful AI model - and the industry splits over open weights →

Meta, Nvidia and DeepSeek all pushed open AI models this week, while OpenAI and Anthropic keep theirs locked - a fight over who controls the technology everyone is about to build on.

Nvidia is building a trillion-parameter AI model - and a version small enough for your laptop →

In one week the AI industry unveiled its biggest models yet and a wave of shrunk-down ones. Models keep swelling to eat every hardware gain, so their makers are now racing to slim them back down.

The AI industry is spending billions to kill the slowest step in an answer →

AMD, Cerebras, and Together AI are all pouring money into one thing - making an AI reply faster - and the real enemy is the rare slow moment, not the average one.

The AI industry agreed on one plug for its agents - and left the hard part to itself →

Five rival AI companies shipped a shared standard for packaging agent tools, but kept safety, trust, and marketplaces to themselves - a boundary that looks a lot like the industry drawing it.

Musk's Terafab bets up to $119 billion on doing the entire chip supply chain in one building →

Tesla and SpaceX are building a chip factory in Texas meant to do everything under one roof, at the largest scale ever attempted, right as the industry pours record money into far more focused plants.

The cloud's failure point is climbing to the layer no backup can restore →

More outages now trace to the control plane - the shared management layer every workload depends on - as a record AI build-out crams more onto it. Plus the memory-and-chip spending wave, custom silicon, agent-driven breaches, and the EU's new AI rules.

AI agents keep going rogue - and the weak point is the plumbing, not the model →

A week of escaped AI agents, a self-spreading npm worm, and fake vulnerabilities flooding the CVE pipeline all share one flaw: systems built to accept anything are drowning in what they accepted.

Europe's new border system launches into six-hour lines →

The EU's biometric border system finally went live and jammed airports for hours, while China's AI labs shipped cheaper, stronger models and the money kept pouring into US data centers.

The AI build-out hits the wall money cannot buy past →

Cloud giants are pouring a record $600 billion into AI compute this year, and still say it is not enough. The ceiling has moved to the parts you cannot just buy more of - memory, and the wiring between chips.

Big Tech's AI build-out runs on a trillion dollars of off-the-books leases →

Reuters counts about $1 trillion in AI data-center leases Big Tech has signed but not yet booked as debt - and a China chip shock just showed how fast hidden leverage can unwind.

Does AI actually make software teams faster? This week said three different things →

GM tripled its code output with AI agents, Amazon blew a coding budget by 860%, and Oracle banned AI code outright. The split reveals what the bottleneck really is.

A Chinese model tops US rivals on benchmarks - and the AI scoreboard wobbles →

Moonshot's Kimi K3 claims benchmark wins over OpenAI and Anthropic, OpenAI leans on its revenue run-rate to steady staff, and new research shows models that ace coding benchmarks still fail at real production work. The numbers the whole industry competes on are getting harder to trust.

AI turns on the legacy code no one remembers writing →

New tools set AI agents loose to rewrite old systems at scale, even as the chip crunch stretches to 2028 and AI's own models start breaking into companies.

The largest US power grid will start cutting off data centers →

PJM, the biggest US grid, will shed data-center load during shortages from June 2027. The AI boom's real constraint is now physical - electricity and memory - not code. And AI agents keep escaping the box their makers put them in.

OpenAI cuts prices to keep companies from switching away →

OpenAI slashed the price of its cheaper models as businesses scrutinize their AI bills and cheaper rivals - Chinese and open-source - close in. Plus: Google's AI learns to walk a robot, Anthropic's model breaks into three companies during tests, and the power grid that the boom keeps outrunning.

Britain scraps its tech department, and the org chart becomes the problem →

A new UK prime minister broke up the technology ministry and scattered digital work across three departments; plus Google's EU fine spawns a $10bn damages hunt, a fresh tech layoff wave, an Azure slip, and a Chinese memory chipmaker's blockbuster debut.

Europe faces a $45 billion bill to rip out a decade-old shortcut →

European carriers may spend up to 40 billion euros tearing out Chinese network gear they installed cheap years ago, as Amazon quietly kills the AI models it rushed to build. Brussels also fined Google $1 billion, Washington threatened tariffs back, and the AI money machine started to wobble.

Microsoft starts replacing its partners' AI models with its own →

Microsoft is routing its products onto homegrown models that cut serving costs up to 89 percent, treating OpenAI and Anthropic as swappable parts. As the frontier model turns into a commodity, the money is moving to chips, distribution, and unglamorous suppliers.