AI Safety news and analysis on Web Pulse: 8280 curated articles tracking the latest AI Safety developments, tools, and research, updated continuously from vetted sources.
Illinois Governor JB Pritzker signed the Artificial Intelligence Safety Measures Act into law, requiring large AI developers like OpenAI and Anthropic to disclose safety practices, report incidents, and undergo independe…
The U.S. Cybersecurity and Infrastructure Security Agency (CISA) is using Anthropic's Mythos tool to audit government code for vulnerabilities, despite a federal ban on Anthropic's AI. The tool has already identified tho…
UK Foreign Secretary Yvette Cooper warned that governments risk an 'AI Hiroshima' if they fail to establish global agreements on AI safety, comparing the challenge to nuclear safety efforts after World War II. She called…
Anthropic's global workspace paper proposes that language models use a cognitive space in the residual stream to represent intermediate reasoning steps as directions, analogous to working memory. The reviewer argues this…
Anthropic published research on July 6th claiming that its AI model Claude has a hidden internal "J-space" that functions as a global workspace for silent reasoning, allowing researchers to read concepts the model uses b…
Reddit is using AI to combat spam and harmful content, reporting it blocks 23 million spam views daily and reduced user exposure to spam by 20% from January to March. The company employs large language models to detect c…
Partnership on AI launched a Global AI Progress Hub and annual Global Responsible AI: Measures of Progress report on July 6, 2026, in Geneva, providing a public platform for organizations to document and compare concrete…
President Trump signed Executive Order 14409 on June 2, 2026, asking frontier AI developers to provide the federal government early access to covered models up to 30 days before broader release. The voluntary framework i…
UN Secretary-General António Guterres opened the first Global Dialogue on AI Governance in Geneva, warning that AI deployment is outpacing governance and that humanity cannot 'vibe code' its future. He called for an inte…
Utah became the first US state to allow an AI chatbot, Doctronic, to renew prescriptions without a doctor through a regulatory sandbox that waives licensing laws. The state's medical licensing board, blindsided by the Ja…
Vercel CEO Guillermo Rauch discussed the company's role in AI software, noting that half of its 6 million daily deployments are triggered by coding agents and over 1 trillion tokens flow through its AI gateway daily. He …
Munich Regional Court I ruled on May 28, 2026, that Google's AI Overviews can be treated as Google's own statements, issuing a temporary injunction in case 26 O 869/26 over false claims about two Munich publishers. The c…
Illinois Governor JB Pritzker signed SB 315, a law requiring large AI developers with over $500 million in revenue to disclose safety frameworks, undergo third-party evaluations, and report critical incidents, with fines…
Apple has added a new popup in iOS 27 and iOS 26 that notifies users and requests permission when their data is sent to Google Cloud to power certain AI features, such as shape generation in iWork and Freeform. The move …
Researchers at Hong Kong University of Science and Technology introduced SkillCloak, a framework that demonstrates how malicious third-party coding-agent skills can evade static install-time scanners through semantics-pr…
Illinois Gov. JB Pritzker signed the Artificial Intelligence Safety Measures Act, making Illinois the first U.S. state to require third-party audits of safety plans for the largest AI labs. The law aims to protect reside…
Anthropic's Fable 5 and Mythos 5 AI models are back online after a U.S. government export ban was lifted, but the lawsuit filed by customer Legion LegalTech Corp over the ban continues. The case challenges the government…
The Model Context Protocol (MCP) will become stateless on July 28, 2026, removing session IDs and the initialize handshake. This change breaks audit trails that relied on session IDs as correlation keys, forcing teams to…
Anthropic secretly embedded a tracker in its Claude Code tool to monitor users in China, using 'prompt steganography' to hide code that flagged timezone, proxy, and potential connections to Chinese AI labs. The company r…
Security firm Sysdig documented the first fully agentic ransomware attack, dubbed JadePuffer, in which an AI agent planned and executed an entire database-extortion operation with no human at the keyboard. The agent expl…