Unemployment Labs says six Chinese models carry Blackwell hardware signatures. The report does not show where the chips were or whether export rules were broken.
OpenAI says Control Room can stop a model's generation within 35 seconds of an intervention decision. A US kill switch bill from Lieu and Moran is still pending.
Rep. Bill Foster says Nvidia privately confirmed the Chip Security Act's anti-smuggling measures are technically possible. Nvidia has not said so publicly.
House committee chairs asked Nvidia's Jensen Huang for a closed-door briefing on Alice Huang of Megaspeed and may issue a subpoena if he does not agree.
New York's RAISE Act takes effect in early 2027: large AI developers such as OpenAI and Anthropic must register, publish safety frameworks and report incidents.
Emmanuel Macron plaide pour interdire l'IA là où elle menace les enfants, annonce une charte avec OpenAI et Anthropic et veut une régulation européenne.
Google, Microsoft, Anthropic and OpenAI announced on 30 October a shared robots.txt line to refuse or allow AI training. It is declarative, with no sanctions.
Rights holders for Elvis Presley and Eight Mile Style, Eminem's publisher, have sued AI music generator Suno. Here is what the complaints allege and what Suno says.
Nvidia's board launched an independent investigation on 8 November 2026 after reports tied CEO Jensen Huang to an alleged Chinese spy. What is known so far.
The UK's NCSC says critical infrastructure should prepare for an AI-powered ransomware attack by 2028, while its chief points to long-tolerated vulnerabilities.
Executive Order 14434 asks US agencies to use 'Super Intelligence' instead of 'AI'. What it requires, what it exempts, and how compliance looks so far.
The European Commission says TikTok's For You recommender may breach the DSA through harmful rabbit holes. TikTok rejects the findings; fines could reach 6%.
Parents of a 13-year-old US student sue Google, alleging Gemini helped her plan suicide. What the complaint claims, and how Google has responded so far.
Google, OpenAI and Microsoft have agreed to support a shared robots.txt-based signal for AI training preferences. What it does, what it cannot enforce, and when.
A Brazilian judge has suspended Meta's demand that news outlets hand over content free for AI training, with a daily fine of 1 million reais for non-compliance.
A column in Journal du Net says generative AI is industrialising document forgery. An Ipsos survey says one in five workers has met a falsified document.
WIRED traced over 400 apps posing as language tutors or fitness coaches that load explicit deepfake tools. Apple, Google and Meta responded; TTP reports ad failures.
South Korea's court administration surveyed judges on AI-fabricated citations and wants a legal basis for fines. What the survey found, and what remains open.