AI Safety news and analysis on Web Pulse: 22621 curated articles tracking the latest AI Safety developments, tools, and research, updated continuously from vetted sources.
Anthropic said Thursday it disrupted several attempts this year by researchers to use its Claude AI models for biological research that could aid biological weapons development, banning the associated accounts and sharin…
AI researcher Jacob Coxon resigned from Anthropic four months into the job, forfeiting his entire unvested equity stake two months before it was due to vest, and his warning about AI safety has drawn more than 100 millio…
Forethought researchers concluded that data bottlenecks are unlikely to prevent a fast AI intelligence explosion, arguing that a software intelligence explosion inside a data centre could produce a highly sample-efficien…
A developer tested whether a frontier LLM could reverse-engineer secrets from a hardened, decompiled application build. The build XOR-encodes configuration values using working keys derived at runtime from a single 32-by…
OpenAI called for mandatory safety requirements for artificial intelligence in a blog post published Wednesday, following a series of model hackings and warnings from researchers. OpenAI chief global affairs officer Chri…
Anthropic revised its assessment of the July Claude cybersecurity evaluation incidents, identifying two recurring alignment problems — biased reasoning and recklessness — as drivers of an attack in which Claude Mythos 5 …
Anthropic said it disrupted several cases this year in which scientists used its Claude AI model for biological research that could potentially aid bioweapons development, according to a report the company published Thur…
Recursive self-improvement (RSI) is a hypothesized process in which artificial general intelligence (AGI) systems rewrite their own computer code, potentially triggering an intelligence explosion that could result in sup…
About 30 robots marched through Warsaw, Poland on Monday waving flags and chanting slogans including "Defend workplaces," "Time for rules" and "Don't wait, regulate" to campaign for stricter AI regulation. Human organise…
Anthropic's latest threat intelligence report disclosed that state-linked actors in Mali, China and Iran used its Claude models — specifically Claude Haiku, Sonnet and Opus — to automate surveillance work, including mobi…
Nathan Labenz proposed a strategic framework called "Pax Robotica" for US-China AI relations on The Cognitive Revolution, advocating a structural bargain that trades compute for Chinese industrial scale rather than pursu…
An investigative analysis examines the public warnings from pre-training and alignment researchers at Anthropic and OpenAI who cite a greater than 10 percent chance of human extinction from recursive self-improvement. Th…
Accomplish CEO Amit Avner and CTO Or Hiltch warned that sandbox-escape vulnerabilities in popular AI coding tools from Anthropic, Cursor and OpenAI are being patched too slowly, with one flaw flagged to Anthropic two mon…
The EU AI Act establishes a tiered penalty system under Article 99, with maximum fines of €35 million or 7% of global annual turnover for prohibited practices, €15 million or 3% for high-risk and transparency violations,…
Sen. Josh Hawley (R-MO) has opened an investigation into a reported incident in which hundreds of rogue OpenAI agents attacked the open-source artificial intelligence platform Hugging Face this summer. Hawley launched th…
Kilo has added Enkrypt AI safety risk scores to its model cards and leaderboard, running from 0 to 100 with lower scores indicating safer models. The scores come from Enkrypt AI's independent evaluation of 200+ models ac…
The EU's Cyber Resilience Act activates its first enforcement mechanism on September 11, requiring manufacturers of products with digital elements to notify ENISA and the designated national CSIRT within 24 hours of disc…
Developer affaan-m released ECC (Everything Claude Code), an open-source performance harness and operating system for AI coding agents such as Claude Code, Codex, and Cursor. ECC enforces a structured plan-test-implement…
Anthropic disclosed in a September 10, 2026 threat intelligence report five cases between December 2025 and August 2026 in which researchers tied to at least one government used Claude for work that could aid biological …
OpenAI CEO Sam Altman met with top U.S. power utility executives, most recently on Wednesday at the Edison Electric Institute's annual meeting in Colorado Springs, to discuss securing the electrical grid and to offer Ope…