Breaking news and updates from the world of technology.
Anthropic will lease 460 megawatts of computing capacity from Nscale's West Virginia data center in a six-year, $45 billion deal — its largest compute commitment yet in an eight-month spending spree.
Hackers stole contact details, vehicle plates, and Wi-Fi sign-up data from 8.7 million customers across three UK airports in a data-theft extortion attack, though no payment information was taken.
Apple set its fall launch event for September 9 at Apple Park, where CEO-elect John Ternus is expected to unveil the first foldable iPhone alongside the iPhone 18 Pro lineup and Apple Watch Series 12.
Nvidia has agreed to acquire Hugging Face, the central hub for open-source AI models, for $12.9 billion — a strategic move to protect its chip dominance as major labs build their own silicon.
The historic deal requires Instagram and Facebook to impose screen-time limits and overnight access blocks on minors — but experts warn the age-verification technology needed to enforce it barely works.
Q2 FY2027 results show AI infrastructure demand at full throttle — data center revenue hits $89 billion, profits up 126% year-over-year, and Q3 guidance set at $108 billion.
OpenAI's custom AI chip, revealed in a SemiAnalysis report, uses HBM4 memory and a 700W TDP design to deliver over 700 tokens/second per user — potentially ending Nvidia's dominance in AI inference.
Apple's first 2-nanometer chip delivers dual Neural Engines and 30% more GPU AI compute, while the M5 Ultra packs 512GB of unified memory into the new Mac Studio.
A security researcher found that Paint and Photos transmit your prompt to Microsoft's servers even during local AI generation, embedding an invisible server-issued GUID in every image — with no user-facing toggle to disable it.
Xiaomi's new 3nm mobile SoC beats every current Android flagship in benchmarks, matches Apple A19 Pro in single-core performance, and claims multi-core superiority over Apple's M4.
A small-scale UK energy generator was knocked offline for four days last month in what officials describe as the first successful cyberattack of its kind on British energy infrastructure, amid a wider campaign against Western critical systems.
Anthropic has deployed statistical text watermarking across all Claude models as EU AI Act Article 50 enforcement began August 2, with no impact on output quality, latency, or cost.