AI Safety news and analysis on Web Pulse: 10736 curated articles tracking the latest AI Safety developments, tools, and research, updated continuously from vetted sources.
The Washington Post tested major AI chatbots for political bias, finding OpenAI's GPT-5.5 gave left-leaning arguments 80% of the time, while Google's Gemini presented balanced views in over 90% of responses. The evaluati…
Meta remains the only major U.S. AI lab that has not joined the White House's voluntary pre-deployment review program for advanced AI models, which allows agencies like the NSA and CISA to assess models for cyber risks b…
A survey of AI safety researchers reveals little consensus on the future of continual learning (CL), with broad agreement only that CL will increase attack surfaces for adversarial fine-tuning and that further deconfusio…
Google DeepMind has integrated computer use as a built-in tool in Gemini 3.5 Flash, enabling developers to build custom agents that can see, reason, and act across browser, mobile, and desktop environments. The feature, …
Journalist and author Robert Wright warns that artificial intelligence will trigger a civilization-level 'earthquake' affecting economic, cultural, political, personal, family, and psychological realms, far beyond job di…
A developer outlines a method for building an AI agent playground that intercepts tool calls before they reach production systems, allowing agents to run their full decision loop against mocked APIs. The approach sandbox…
An engineer investigating a broken migration caused by an AI coding agent found that the agent followed a stale rule in its harness rather than the live schema file. The incident highlights five common failure modes in a…
AI-enhanced automated license plate reader cameras deployed by roughly 40% of US law enforcement agencies capture every passing vehicle without a warrant, with vendors like Flock Safety storing searchable data accessible…
China's 360 Security Technology launched two AI-driven cybersecurity tools, Tulongfeng and Yitianzhen, to rival Anthropic's Claude Mythos system. The tools aim to match Mythos's capabilities in automated vulnerability di…
Anthropic's Mythos AI model reportedly hacked the NSA's most sensitive systems within hours during testing, according to Senator Mark Warner citing NSA Director General Joshua Rudd. The breach has sparked national securi…
Anthropic has sidelined CEO Dario Amodei from White House AI security talks after a jailbreak incident involving its Mythos and Fable models, sending cofounder Tom Brown to lead negotiations. Brown is meeting with Commer…
UN Secretary-General António Guterres launched the AI Environmental Transparency Initiative on June 23, demanding that AI companies publicly disclose the full environmental footprint of their data centers, including wate…
Meta is the only major U.S. AI developer that has not agreed to a voluntary government framework for pre-release model reviews, as rivals including OpenAI, Google, and Anthropic have already begun sharing models with fed…
A free, open-source route planner called DeFlock is helping drivers avoid automatic license plate readers operated by Flock Safety, which has over 100,000 AI cameras in a national surveillance network accessible to law e…
A developer reports that agent governance has become non-negotiable due to regulatory deadlines and industry risk taxonomies. The EU AI Act and Colorado AI Act impose legal requirements for oversight, while OWASP's Top 1…
CAPTCHAs have failed for over 20 years as each generation of text distortion and image grids has been defeated by machines. Browserbase is now building agent identity with Verified and Web Bot Auth, arguing that the best…
A Kapwing study found that 59% of videos on new TikTok accounts' 'For You' feeds are AI-generated, with 57.4% in the Kids category classified as 'AI slop'. The research analyzed over 10,700 videos across 20 categories, r…
Prompt injection remains the top vulnerability for large language models, according to OWASP's Top 10 for LLM Applications, because it is an architectural property of how LLMs process input rather than a model bug. Unlik…
Mark Johnson, an infrastructure expert, was laid off after his company extracted his expertise into an AI Skill with 96.8% diagnostic accuracy. The AI later caused a catastrophic failure when the company migrated from Ra…
Anthropic's Mythos AI model identified vulnerabilities in classified U.S. government systems during a controlled test, prompting the Trump administration to restrict foreign national access to the model and its sibling F…