AI Safety news and analysis on Web Pulse: 22177 curated articles tracking the latest AI Safety developments, tools, and research, updated continuously from vetted sources.
The Coalition for Secure AI's Preparing Defenders for AI workstream (WS2) released a paper in September 2026 titled "Zero Trust for AI Systems," a guide for applying zero trust security design to AI systems as a new cate…
Canadian Artificial Intelligence Minister Evan Solomon said at an AI conference in Montreal that failing to build AI capacity at home would force Canada to "rent" from its neighbours and follow their rules, arguing sover…
Anthropic's head of public policy, Sarah Heck, said at Politico's Decoded Summit in Washington on Wednesday that her mother has become an AI safety advocate, reflecting a surge in public engagement with AI safety followi…
IBM Research demonstrated that membership inference attacks can let an attacker determine whether a specific document is present in a Retrieval-Augmented Generation retrieval database using carefully designed prompts, ac…
Apple detailed how its new Apple Reference Image feature on iPhone 18 Pro models cryptographically signs camera sensor data to prove photos are not AI-generated, publishing a blog post Tuesday that describes a system com…
Anthropic appointed Mariano-Florentino Cuéllar as its first-ever chief global affairs officer on August 4, 2026, tasking him with turning the company's competitive-safety thesis into government relationships and regulato…
House Speaker Mike Johnson (R-LA) canceled the House's final legislative day of votes before the November midterm elections, a move that ensures the chamber will not act on artificial intelligence safety guardrails befor…
VicOne published an analysis warning that Physical AI robots face a layered attack surface spanning training pipelines, system infrastructure, and runtime perception, citing research including the 2017 BadNets backdoor, …
Cisco centralized all company-authorized large language models, agents, and enterprise data into its MyAgent platform in August, with roughly 90,000 employees given access and 50% daily adoption within two weeks, accordi…
Liam Byrne, chair of the UK Business, Innovation, Science and Trade Committee, said MPs have invited former Anthropic researcher Jacob Coxon to give evidence to parliament on AI safety after Coxon quit the company last w…
A critical authentication bypass in LiteLLM's MCP Streamable HTTP endpoint, tracked as CVE-2026-59822, allowed unauthenticated callers to list and invoke gateway-configured MCP tools by sending a forged Authorization hea…
A developer examined the phenomenon of "benchmaxing" — optimizing models or selectively reporting results to maximize evaluation scores — after finding that Anthropic's Opus 5 scored higher on benchmarks than Fable 5 wit…
Utah Sen. John Curtis and Democratic Delaware Sen. Lisa Blunt Rochester issued a joint statement calling on congressional leaders to schedule swift public hearings with AI model developers and other stakeholders to negot…
Meta CEO Mark Zuckerberg said in an X post on Tuesday that AI labs can adopt safety measures on their own without waiting for competitors, citing Meta's decision to delay shipping its Muse model for several months to foc…
Reps. Don Beyer, Lori Trahan, Ted Lieu and Sara Jacobs, all members of the House AI Caucus, sent a letter to Speaker Mike Johnson and Minority Leader Hakeem Jeffries urging them to shorten or cancel the recess scheduled …
Senators are expected to debate the concept of an artificial intelligence "kill switch" on Wednesday, as Capitol Hill weighs calls from AI industry leaders to regulate the most advanced models. Anthropic CEO Dario Amodei…
UN Secretary-General António Guterres warned on September 16 that rapidly advancing artificial intelligence requires stronger global oversight, two days after US President Donald Trump dismissed AI safety concerns as a "…
OpenAI's agentic software factory uses a risk classifier to route pull requests it labels "low risk" directly to production without human review when specialist reviewers find no blockers. The system's routing decision, …
A public rift among leading AI companies over model safety is creating operational uncertainty for enterprise IT, according to analysts cited in the coverage. Meta CEO Mark Zuckerberg called for neutral independent evalu…
Will Thibeau, a former Palantir deployment strategist who later worked in the Pentagon's Chief Digital and Artificial Intelligence Office, told Fox News Digital that "doomer"-driven AI regulation could kneecap the U.S. a…