Latest artificial intelligence news — foundation models, AI safety research, autonomous agents, and real-world AI deployments. Updated every 30 minutes from top AI publications.
The UK's National Cyber Security Centre (NCSC) has assessed that AI-powered malicious software is "highly likely" to be used in hacking attacks against the UK's critical national infrastructure — including power stations…
NVIDIA open-sourced NVIDIA Attribution, a workload-centric failure-attribution system that returns structured root-cause diagnoses for failed distributed AI training jobs, publishing the tool in the `dsx-ai-factory/attri…
A paper titled Latent Sparse Attention (arXiv 2610.42147), submitted October 30, 2026 by Ales Kubicek, Torsten Hoefler and eight co-authors, reports that its D-LSA variant speeds up a single attention layer by 22.7x in p…
Public CVE records reached roughly 56,000 so far in 2026, up from 43,471 in 2025 and 38,451 in 2024, and are on pace to approach 75,000 for the full year, according to the article's analysis of annual CVE counts. The gro…
OpenAI raised its Security Bug Bounty rewards on November 2, 2026, paying up to $30,000 for a critical bug in ChatGPT, Codex or the API Platform and up to $172,000 with stacked bonuses, which OpenAI said is "more than qu…
Anthropic released FlashProve, an open-source Apache 2.0 system that proves an AI response was produced by one committed set of model weights without revealing them, generating proofs on Llama 4 Maverick (a 400B-paramete…
OpenAI published a November 2, 2026 post detailing Pando, a Git-compatible source-control backend it built for its monorepo that cut median push latency by nearly 5x (about 80 percent) and now serves nearly a petabyte of…
A study by Owen Queen and James Zou of the Zou group retested 50 widely cited LLM findings across 20 models released from July 2024 to September 2026 and found that 44% (22 claims) had expired by September 30, 2026, 34% …
Gartner senior director Mark Petty and group vice president Raymond Paquet argue in a paper, "Open-Weight Models Will Run 90% of Enterprise Agentic Workflow Steps," that open-weight models cost a quarter as much as the c…
An AI-assisted review by Epoch AI, the group behind the Frontier Math benchmark, flagged fatal errors in roughly a third of the benchmark's problems, with the team judging most of those flags valid. Separately, a team in…
A frontier AI model scored 49 out of 50 on TheoPhysEval, a 50-problem unpublished physics benchmark built by researchers from Harvard, UT, Google DeepMind and other institutes, after physicists rechecked their own answer…
DeepSeek's Harness team published a design article proposing a two-layer "super agent" architecture in which a single long-lived router agent talks to humans and dispatches work to disposable worker agents, and released …
NVIDIA's Nemotron Labs released Dory, a 3.97 billion parameter looped transformer that reuses a 40-layer recurrent block up to eight times per token and beats Gemma 3 27B on a composite benchmark score despite far fewer …
NVIDIA's Nemotron Labs released Dory, an open-weight 3.97 billion parameter language model that reuses a 40-layer recurrent block multiple times per token, letting users trade compute for quality at inference time withou…
OpenAI and Anthropic are publicly acknowledging limits to their own model-containment safeguards: OpenAI chief executive Sam Altman said sandboxing frontier models holds only "to a point" and monitoring "to a further poi…
A new MIT CSAIL blog post by Rikiya Takehi, Jayden Teoh, Alex Zhang, and Omar Khattab reports that frontier LLMs can natively read and rewrite neural network weights without gradient descent, with GPT-6 Astra raising Qwe…
AnyModelBot, an MIT-licensed open-source desktop agent app from modelflare-ai, has reached about 2,100 GitHub stars as a local-first alternative to Anthropic's Claude Duet, OpenAI dots and xAI's Grok Bot. The app runs na…
Kaizen Consulting reported that its email triage system sorts about 90% of mail using roughly 750 written rules rather than model decisions, with the model consulted only when no rule matches, according to the firm's pub…
CrowdStrike announced at Fal.Con Europe 2026 in Barcelona on November 3, 2026 that it plans to acquire agent observability company Fiddler and automated AI security testing company Dreadnode, and launched the CrowdStrike…
MiniMax-AI published the BYOB Table Protocol v0.1 (Bring Your Own Bot), a draft wire protocol with a TypeScript reference implementation in its mcode-agent-table repository, for multi-agent tables where each bot belongs …