AI Safety news and analysis on Web Pulse: 12884 curated articles tracking the latest AI Safety developments, tools, and research, updated continuously from vetted sources.
Anthropic sent security researcher Nicholas Carlini to Washington to address US government concerns about AI safety after export control directives forced the company to temporarily suspend global access to its Fable 5 a…
Researchers found that pruned large language models can pass multiple-choice benchmarks but fail to answer the same questions in open generation, creating a 'benchmark illusion.' The study, using multilingual question an…
A LessWrong post proposes applying exploratory modeling, a decision-making methodology used by RAND, to AI governance challenges. The project would build computational models to stress-test policies across diverse plausi…
Cyber extortion group FulcrumSec claimed on Tuesday to have stolen over a terabyte of data from pharmaceutical giant Novo Nordisk and attempted a $25 million extortion. The group said it is now exploring private sales of…
A developer from MartinLoop argues that unbounded retry loops in agent systems are the primary driver of cost and trust erosion, not initial failures. The team implemented a control-layer policy with budget caps, max att…
The U.S. Commerce Department used export control law to shut down Anthropic's most advanced AI models, marking the first time such controls have been enforced against a live AI system. The action, which effectively block…
A researcher who previously warned about existential AI threats now advocates for broader access to cutting-edge systems, reflecting a shift in the AI safety community. The debate intensified after Anthropic's Claude Myt…
Colorado's original AI law, SB 24-205, was judicially stayed and replaced by SB 26-189 after xAI sued and the DOJ intervened. The new law, effective January 1, 2027, eliminates annual impact assessments, risk management …
A project proposal advocates for using computational exploratory modeling and robust decision-making frameworks to improve AI governance preparedness. The approach emphasizes generating thousands of scenarios and designi…
The UK's Department for Science, Innovation and Technology has partnered with BCS, The Chartered Institute for IT, to launch the AI Assurance Stakeholder Consortium, tasked with developing a voluntary professional code o…
A Calgary woman used ChatGPT to self-diagnose diabetic ketoacidosis after a seven-hour emergency room wait in 2026, but a subsequent safety evaluation found OpenAI's ChatGPT Health under-triaged 52% of genuine emergencie…
Educational technology company Chegg faced a sharp stock decline in May 2023 as generative AI tools like ChatGPT disrupted its business model, illustrating how deep expertise can become a blind spot during rapid change. …
New Mexico State Police revealed that a CT scan of the reconstructed skull of missing Los Alamos National Laboratory employee Melissa Casias found no bullet, casting doubt on suicide theories. Former FBI agent Ben Hansen…
The US Justice Department filed to intervene in a Clean Air Act lawsuit against Elon Musk's xAI on June 16, arguing that unpermitted natural gas turbines at its Mississippi data center are essential for national security…
A security service blocked access to an article about AI and government surveillance, preventing the content from being read. The block was triggered by a specific action, such as submitting a certain word or phrase, a S…
The US government has declined to add Chinese AI firm DeepSeek to its trade blacklists despite documented national security concerns, state-level bans, and congressional legislation. Federal officials say there is 'nothi…
G7 leaders discussed a plan to grant select 'trusted partners' access to advanced U.S. AI models from Anthropic, potentially bypassing restrictions on non-American use. The talks, held on the sidelines of the G7 summit i…
OpenAI's ChatGPT will be deployed on the Pentagon's GenAI.mil platform in early July, providing a custom, government-approved version to over 3 million defense personnel. The instance is certified for controlled unclassi…
The Pentagon disclosed in a sworn filing that xAI's Grok chatbot was used to support over 2,000 missile strikes within 96 hours, marking a significant deployment of commercial AI in military operations. The filing, repor…
The Commerce Department imposed export controls on two of Anthropic's advanced AI models on June 13, forcing the company to globally disable its Fable 5 and Mythos 5 models. The move reignited congressional debate over A…