Google DeepMind Demonstrates Watermarking for AI-Generated Proteins
Google DeepMind disclosed SynthID Bio on September 30, a proof of concept for watermarking AI-generated proteins that embeds an imperceptible signature directly into biological code, making the watermark verifiable not only on a digital mod…
OECD HAIP 2.0 Framework Aligns 80% With EU AI Act
A June 2026 analysis by the Centre pour la Sécurité de l'IA (CeSIA) found that 80% of the 31 reporting questions in the OECD's Hiroshima AI Process (HAIP) 2.0 framework refer to requirements covered by the EU AI Act regime, with 58% showing…
Nvidia Opens 2027–2028 Graduate Fellowship Applications, Up to $60K
Nvidia Corp. opened applications for its 2027–2028 Graduate Fellowship Program, offering awards of up to $60,000 to doctoral students conducting research relevant to Nvidia technologies . Now in its 26th year, the program provides grants, m…
White House AI accord bets on self-policing as Congress stalls
The White House's new voluntary AI accord, signed by six frontier-lab chiefs and the president, arrives on the same day a Senate Republican blocked the only live AI-regulation bill on the floor. The juxtaposition defines the actual state o…
ALL BRIEFS
RANKED · 9 THIS DAYOpenAI's DevDay stack bets ChatGPT can become the workplace OS
OpenAI shipped Dots alongside safety disclosures and a scrapped model launch, framing its most ambitious DevDay yet as a platform bet that treats ChatGPT not as a chatbot but as the control surface for knowledge work. The juxtaposition of…
FTC compels OpenAI, Anthropic to hand over safety records in rogue-agent probe
The Federal Trade Commission is expanding a sweeping investigation into OpenAI, Anthropic, and other frontier AI labs over the potential dangers their products pose to consumers . FTC Chair Andrew Ferguson is preparing civil investigative d…
Anthropic flags GLM-5.3 as exploit-capable frontier model released without effective safeguards
GLM-5.3, the open-weight model from Zhipu AI (known outside China as Z.ai), almost matches Anthropic's most advanced model in cyber capabilities but carries far weaker safeguards, Anthropic warned in a report from its Frontier Red Team . Th…
DeepSeek open-sources six Ascend-native tools to cut Nvidia out of Huawei's AI stack
Deep Seek has released six open-source software modules built for Huawei Technologies' Ascend AI chips, the two Chinese firms disclosed on Wednesday . The tools include TileLang, positioned as an open-source alternative to CUDA, Nvidia's pr…
Google's AI publisher-payment pilot draws frustration as most payouts stay marginal
Google's pilot program paying roughly 100 digital publishers for content used in its AI search features is generating frustration among participants, with several small and mid-sized publishers reporting payouts that equal roughly one-tenth…