The AI knowledge 99.9% of people never discover
LLM hidden states hold richer confidence than models verbalise; a new method extracts it before generation finishes, affecting any AI system needing to know whe
A new arXiv preprint reframes neural network training as optimal control, inserting layers where error is highest — but evidence is limited to scientific datase
OpenAI's new analysis exposes reliability issues in SWE-bench Pro — the very benchmark it recommended to replace the contaminated SWE-bench Verified — leaving A
OpenAI Academy and the Walton Family Foundation are launching hands-on AI Skills Jams to help K-12 educators build practical classroom AI skills.
Anthropic researchers found a small broadcast hub of silent concepts inside Claude — the J-space — that drives its deliberate reasoning, and built a lens that lets them watch it think.
An ETH Zurich preprint proves four discrete diffusion methods optimize the same object — and reveals why one popular parameterization diverges at initialization
GaP, a multi-agent coding framework on arXiv, builds computation graphs for robot tasks and rehearses them in simulation to tackle variable automation's reliabi
A new arXiv preprint from UC Berkeley argues that language models measuring culture have already absorbed it — making neutral measurement impossible.
A new arXiv preprint distils 25 years of cybersecurity incident response research into a single taxonomy, comparing it against NIST and seven other frameworks —
