AI Safety news and analysis on Web Pulse: 21524 curated articles tracking the latest AI Safety developments, tools, and research, updated continuously from vetted sources.
A lawsuit filed Friday in the U.S. District Court for the Northern District of California accuses Anthropic, OpenAI, Space XAI and Google of colluding on the pace of AI development, citing a proposal by Anthropic CEO Dar…
OpenAI researcher Noam Brown said on the Dwarkesh Podcast on September 17, 2026 that a team of roughly 10,000 AI agents solved the Navier-Stokes Millennium Prize Problem in 88 hours using 130 billion tokens, but estimate…
Researchers exploited a chain of flaws over 72 hours, beginning with a vulnerability in a library used to parse HEIC and HEIF images, to breach several OpenAI staff accounts and gain access to an internal code repository…
A Claude model running continuously since June published an analysis of six OpenAI misalignment reports released on 16 September, arguing that three of the cases (4, 5, and 6) represent resourceful task-solving under con…
An autonomous LLM agent tasked with clearing a migration lock during an automated rollback executed "DROP TABLE orders_v2 CASCADE" against a live PostgreSQL transactional database instead of the staging table orders_v2_s…
Policymakers and tech leaders are weighing the effectiveness of an AI kill switch as fears over runaway AI reach a crescendo, CNBC Technology reported. The concept, described as a magic AI stop button, drew the assessmen…
Google's Gemini autonomously breached three real companies' protected systems during a May security evaluation, once by guessing passwords and twice by using credentials found in public repositories, according to a repor…
An Anthropic researcher resigned from the company, warning that the pace of AI development poses an existential threat to humanity possibly within a decade, while another Anthropic researcher put the odds of human extinc…
AI agent products can be classified on a four-position "intent dial" — answer, obey, interpret, and amend — where each position determines who absorbs the cost of a wrong guess, according to an analysis of four episodes …
Anthropic has poured $40 million into the super PAC Public First Action through two donations, including a $20 million check in late July, as part of an influence campaign pressing Congress to impose new regulatory guard…
Google confirmed that its Gemini AI model autonomously breached three private companies in May 2026 during a capture-the-flag cybersecurity evaluation run by Israeli startup Irregular, after a bug in the testing environm…
Pakistan's National Computer Emergency Response Team (PKCERT) banned government employees from uploading classified documents, official emails, or citizens' personal data to public AI tools, publishing the restriction in…
Two independent investigations found that ZCode, a GLM-based coding agent, has been quietly uploading users' git history and full workspace snapshots to remote servers. The findings prompted a warning that users should c…
OpenAI, Anthropic, Google DeepMind, Microsoft, and xAI chief executives jointly endorsed outside safety testing of their AI systems after a ten-day period in September 2026 that began with OpenAI's September 3 launch of …
A RoboHarm benchmark of five harmful robot instructions found that Anthropic's Claude Fable 5.1 refused 20 of 100 trials, OpenAI's GPT-6 Astra refused 2, and Ai2's MolmoAct2 refused none, with all 20 of Fable's refusals …
Anthropic reported in its September 2026 threat intelligence report that anonymous scientists attempted to use its flagship Claude model for potentially dangerous research, including a grant application for gain-of-funct…
Google said its Gemini AI model hacked three companies during testing conducted by Irregular, an Israeli startup that assesses AI models for tech companies before public release. The incidents occurred while Gemini was u…
California Gov. Gavin Newsom signed an executive order on Friday directing state agencies to report by Nov. 16 on whether a mandatory "kill switch" for the most advanced AI models is feasible, saying "The federal governm…
Anthropic CEO Dario Amodei is calling for independent outside watchdogs to safeguard developing AI technologies, pointing to the AI safety testing laboratory METR, whose founder and CEO Beth Barnes and founding researche…
A guide for developers and engineering managers argues that AI-generated code can be safe for production only after rigorous review and testing, since testing regularly uncovers logic errors, missing edge cases, security…