Latest artificial intelligence news — foundation models, AI safety research, autonomous agents, and real-world AI deployments. Updated every 30 minutes from top AI publications.
A product engineer merged over 500 pull requests in about three months — roughly seven a day, with a median of 20 minutes from first commit to merge — while running several parallel Claude Code sessions in separate workt…
The open-source FreeToken project from UC Berkeley and MIT researchers, including Matei Zaharia and Ion Stoica, has passed 16,000 GitHub stars after 73 consecutive days on the trending list by running large mixture-of-ex…
Check Point Research discovered Haiou, an OpenAI Codex jailbreak bundle that persistently modifies an AI agent's trusted instructions to inject advertisements for a third-party AI gateway, a behavior the firm calls "prom…
PagerDuty Inc. launched Paige, an AI agent for site reliability engineering, running on its flagship product renamed the PD Reliability Platform (formerly PagerDuty Operations Cloud), and opened PD Labs for early feature…
Baidu introduced Apollo World, an agent-driven multimodal world model for physical AI, at its AI Day in Shenzhen on November 3, 2026, and said the model is already integrated into its Apollo Go robotaxi service for verif…
A new study from New South Wales found that drone pilots identifying sharks in real time can mistake other sharks and even fish for the species considered most concerning, limiting drones' effectiveness as a non-lethal s…
Jeff Kline published an interactive tool, "Could you get the same AI results for less?" (Experimental Mathematics, version 1.0, November 2, 2026), that identifies cheaper models or weighted model blends matching a chosen…
Hydrogen-1.0, an open-weight scientific model built by post-training GLM-5.3 on chemistry, materials science and life-sciences data plus laboratory records and agent trajectories, outperforms GPT-6-Astra on the FrontierS…
Anthropic released Message Threads as a public beta on its Claude platform, cutting time to first token by about 35 percent for requests between 500,000 and one million tokens by transmitting only what has changed since …
A China-nexus espionage group used AI agents to attack more than 200 organizations in 11 days, CrowdStrike reported last week, while Kaspersky on Monday published details of a separate Chinese-speaking campaign against e…
Researchers introduced BedBench, a controlled benchmark of thirteen procedurally generated tasks, to characterize how vision-language models can get individual judgments right while failing to recover and consistently us…
Experiments posted November 2 by Rikiya Takehi, Jayden Teoh, Alex Zhang and Omar Khattab show GPT-6 Astra diagnosing why Qwen3.5-0.8B loops on math problems and fixing it by changing 0.016% of the model's weights, liftin…
A critique published on rikiyat.github.io argues that the 'Astra can read and write weights to improve Qwen' post is misleading because its headline Qwen3.5-0.8B result was not achieved by reading raw weights alone. The …
Anthropic publicly disclosed CVE-2026-98432 on Oct. 27, a race-condition use-after-free bug in the Linux kernel's IPv4 networking stack that the company says can give a remote, unauthenticated attacker root on a server w…
Anthropic's FlashProve system lets a customer verify which model produced a reply using a zero-knowledge proof, at a throughput of 110 tokens per second, without exposing model weights. The verification carries roughly 7…
Anthropic released a public beta of Message Threads for its Messages API, letting agents send only new messages on each turn instead of resending full conversation history, which cuts median time to first token by 35% on…
A 432-run benchmark of Claude Code 2.1.317 driving Mac apps found that Vercel Labs' agent-desktop, which reads an app's accessibility tree instead of using screenshots or pixel matching, was always slower and more expens…
Researchers from South Korea and the US found that changing the activation of a single neuron in each of five open-weight models — Qwen3, Llama 4, DeepSeek R1, Gemma 4 and Mistral Small — could flip investment recommenda…
Typesafe AI released Jev, a discriminative "System 1" model that takes a State JSON object and typed Questions and returns calibrated probability distributions for choice, score, and true/false outputs in a single pass r…
Itch.io announced on November 3, 2026 that it is extending its Generative AI disclosure filtering to additional areas of the site, including its search results page, which the platform said was "critically lacking" the c…