The Financial Times reports Nvidia is ordering data-centre customers to cut their China exposure, backed by site inspections and end-user checks; Nvidia has not commented publicly.
In a blog post the authors call very early results, GPT-6 Astra read raw weights, traced Qwen's reasoning loops and patched them by hand, lifting MATH-500 from 7.9% to 48.1% and IFEval 3x; it is not peer-reviewed.
In case C-250/25, Like Company v Google, Advocate General Szpunar advised that a chatbot retrieving and reusing a press article to answer a user may infringe the press publisher's right; the opinion is not binding and the Court has yet to rule.
The report says the packages are classed as server processors to get around export rules and the memory is desoldered in China; it is a single press report and the companies have not commented in the coverage.
Reports say the chips will sit in a new data centre in Ulanqab, Inner Mongolia, with pre-training starting as soon as late 2026 and Huawei engineers embedded with DeepSeek's team; neither company has confirmed.
Beijing announced new central bank swap operations on Monday, with 31 brokers and fund managers applying for more than 100 billion yuan, and urged tech firms to buy back shares; the Shanghai Composite fell 1% on Tuesday.
The S-1/A says internal estimates about future models, not current ones, point to its Responsible Scaling Policy's line for "dramatically accelerated progress attributable to AI"; no current model has been designated as crossing it.
The benchmark closed at 5,648.79 with SK hynix down 5.03% and Samsung Electronics down 4.15%, after the PHLX chip index fell 3.4% overnight; analysts cited doubts about returns on AI infrastructure spending.
Saab says its agents, trained over millions of simulated flight hours in its AI Fighter Pilot Academy, have validated beyond-visual-range air combat in real flight; when they enter service depends on Sweden's customers.
The zkSNARK prototype proves Llama 4 Maverick outputs at about 110 tokens a second on one H100, roughly 70 to 95 times the GPU time of normal inference, with code and a technical report released.
The NCSC's State of the Cyber Nation report, published Tuesday, says hostile actors' use of AI will highly likely raise the scale and efficiency of attacks within two years; chief executive Richard Horne says the era of "tech on the cheap" is over.
Cathedral researchers will use OpenAI's models alongside its teams to attack its defences; a critical bug in ChatGPT, Codex or the API now pays up to $30,000, and the sprint's start date is not given.