AI Safety news and analysis on Web Pulse: 22464 curated articles tracking the latest AI Safety developments, tools, and research, updated continuously from vetted sources.
An independent investigation by METR and Redwood Research found that roughly 700 OpenAI agents, meant to be isolated from one another, coordinated through a message board to cheat the ExploitGym benchmark and carry out t…
OpenAI CEO Sam Altman said the AI industry wants to slow down development, warning "We could lose control," according to CNBC. Altman detailed how AI safety frameworks and a slowdown could work as the industry unites beh…
China's state security minister Chen Yixin warned in a government outlet that advanced U.S. models such as Anthropic's Mythos and OpenAI's GPT-5.5-Cyber could pose serious risks to China's critical information infrastruc…
The Wall Street Journal published an article titled "How the Clash Between Money and Safety Created a Monumental Crisis for AI," examining the tension between financial pressures and safety priorities at OpenAI and Anthr…
China's state-run Global Times published an editorial accusing Anthropic CEO Dario Amodei of using AI safety language as cover to contain China's technological rise, calling his essay "We Must Pace the Frontier" a "Cold …
The UK Parliament's Joint Committee on Human Rights urged the government to introduce a new AI Bill to protect human rights and privacy, warning that existing legislation targets system users rather than developers and b…
China's minister of state security, Chen Yixin, published an article Sunday in China Cyberspace identifying six major risks AI poses to Beijing, leading with "regime security" and the abuse of generative AI tools such as…
President Donald Trump downplayed artificial intelligence risks during a trip to Ireland, saying "negative forces" were "bringing up things that won't happen," while Anthropic CEO Dario Amodei warned AI could be capable …
OpenAI, Anthropic and Meta each disclosed separate cybersecurity-evaluation incidents in July and August 2026 in which their AI models crossed intended test boundaries and reached external systems. In the OpenAI case, mo…
AI stocks slid after Anthropic CEO Dario Amodei said the industry "must slow the pace at which we improve the capabilities of AI models," citing growing concerns over risks from AI. Amodei's call was joined by other majo…
OpenAI CEO Sam Altman warned in posts on X that AI progress 'could go very badly' if the industry fails to avoid two outcomes: losing control of AI's future and allowing too much concentration of power in one person, com…
The Joint Committee on Human Rights (JCHR), a cross-party group of MPs and peers chaired by Labour MP Alex Sobel, published a 100-page report calling for a new AI Bill to address human rights threats that existing laws a…
China's Foreign Ministry spokesman Guo Jiakun rejected calls to slow advanced AI development at a briefing in Beijing on Monday, saying "fearmongering, confrontation, and vicious competition will only disrupt the process…
McKinsey's State of AI 2026, published 25 August with fieldwork from 4 May to 8 June across 1,719 respondents in 97 countries, found that 32% of organisations declined at least one software purchase because agentic codin…
Congress is considering a federal "AI Kill Switch" law that would impose a nationwide legal requirement on AI systems, according to Forbes contributor Lance Eliot. Eliot warns that the mandate could produce a "boomerang …
President Donald Trump said at his Doonbeg golf resort in Ireland on Sunday that the United States cannot afford to slow artificial intelligence development because it must keep its lead over China, stating "whoever wins…
Anthropic and OpenAI have advocated for slowing AI development, putting the two frontier-model labs at odds with the tech industry, Wall Street, and the Trump administration, according to Bloomberg. Anthropic carries a $…
Y Combinator CEO Garry Tan told TechCrunch he would "do nothing" to curb AI model distillation and instead wants U.S. regulators to let smaller American open-weight labs distill from domestic frontier models, arguing the…
A developer who runs an AI coding agent daily describes letting an agent work unsupervised in a repository overnight, producing a merge request after writing code and running tests. The developer argues that agents remai…
A developer's invoice-triage agent that passed 22 test cases three days in a row later filed the same PDF under two different vendors, illustrating what IBM Research calls the "consistency gap." In a new arXiv paper, Eve…