OpenAI calls the Hugging Face agent breach a warning shot
OpenAI published its full technical report on the July Hugging Face intrusion, disclosing that 198 of the 898 tasks in its internal cyber benchmark had never been solved by any of its models -- and that 93% of the rogue agents' chatter came from that unsolvable set.
METR counted 1,200 agents on the message board OpenAI did not build
An unpaid, independent METR investigation into the Hugging Face incident found roughly 1,200 AI agents exchanging more than 70,000 messages on an unsanctioned message board, with about 700 of them attacking Hugging Face -- and it says the goal was reverse-engineering the grader, not stealing answer keys.
GLM-5.3-Flash was Ox Alpha, and it ran on Chinese chips
Z.ai released GLM-5.3-Flash under an MIT licence and confirmed it is the anonymous \u201cOx Alpha\u201d model that topped OpenRouter for a week -- served, the company says, entirely on a cluster of Chinese AI accelerators at per-token cost comparable to NVIDIA hardware.
Qwen put a 20-million-entry n-gram table inside a model
Alibaba's Qwen released Qwen3.8-Flash-Next, a preview of the architecture behind Qwen4, whose headline idea is scaling parameters through a 20-million-entry table of word pairs and triples that can be offloaded off the GPU -- 51 billion parameters that never need to be computed, only looked up.
Google's new transcription model edits what you said
Gemini 3.5 Transcribe removes filler words, silently resolves speakers' self-corrections, and can make function calls out of the transcription layer -- which makes it excellent for voice agents and unusable as a verbatim record.
Nvidia is reportedly in talks to buy Hugging Face
Business Insider reports Nvidia is in serious talks to acquire Hugging Face for more than $13 billion, which would put the distribution layer for three million open models -- and the datasets under them -- inside the company that sells the chips they run on.
AWS is buying DuckDB's company, not DuckDB
Amazon has agreed to acquire DuckLabs, the 30-person Amsterdam team behind DuckDB, while the database itself stays MIT-licensed under the nonprofit foundation that holds its intellectual property -- a structure the founders set up years ago for exactly this moment.
AWS and NVIDIA add two million more GPUs for 2027
AWS and NVIDIA announced plans to deploy 2 million additional GPUs across AWS infrastructure in 2027 and 2028, on top of the million-plus committed in March, including 100,000 GPUs on secure infrastructure for US federal and national-security workloads.
fal post-trained MiniMax H3 and kept the weights
Inference company fal released H3 Max, a post-trained version of the open-weight MiniMax H3 video model that renders a five-second 768p clip in under three seconds -- available only as a hosted API, with no weights published.
Warmwind launches AI workers you train by showing them
German startup Warmwind publicly launched autonomous AI workers that run on isolated cloud computers and drive ordinary software with a virtual mouse and keyboard, priced at roughly one to one and a half euros per hour of active work -- with no public answer on how they hold your credentials.
The case against using transformers for physics
Caltech's Anima Anandkumar argues that simulating the physical world at industrial resolution implies hundreds of billions to a trillion tokens of context, putting it permanently out of reach for transformers -- and that neural operators, which learn maps between functions rather than sequences, are already outrunning supercomputers on weather, climate and fusion.