77 stories from 16 sources
The week in AI and Technology
Click on a headline to see stories that shaped it.
Threads
6 threads
The week connected a broader political fight over AI rules with direct government influence over frontier-model releases. Officials shaped the timing of GPT-5.6 and the terms for restoring Mythos 5 access, making regulatory review an operational constraint.
China appeared repeatedly in the computing race, with LineShine taking or claiming the top supercomputer position despite export controls. Restricted Nvidia chips became more expensive, while Asian startups moved to fill gaps with local frontier-style models.
The AI buildout moved from large spending commitments toward a contest for scarce capacity, including GPUs, data centers, power, memory, and model-serving access. Providers also began pursuing custom chips and passing infrastructure costs into hardware prices.
Security concerns expanded from intelligence warnings about AI-enabled cyber operations to accusations that Alibaba used fraudulent accounts and millions of queries to extract Claude's capabilities. The developments put both model misuse and model theft under sharper scrutiny.
European firms showed broad but shallow AI use, with intensive adoption remaining rare. Concerns about missing return-on-investment metrics added pressure to the next phase of enterprise deployment.
Figma's AI additions progressed from motion-graphics and shader tools toward a full-stack canvas for end-to-end product creation. The shift shows software companies using AI to expand beyond their original workflow niches.
Day’s Recap