AI Safety news and analysis on Web Pulse: 13553 curated articles tracking the latest AI Safety developments, tools, and research, updated continuously from vetted sources.
A giant inflatable caricature of Elon Musk appeared in Times Square on Thursday, one day before SpaceX's record-setting initial public offering, as advocacy group Safe AI Now protested the company's AI platform Grok for …
AOL cofounder and Revolution CEO Steve Case described artificial intelligence as a "huge, huge opportunity" but warned it will likely be a "net negative" for jobs, speaking at Fortune's Brainstorm Tech conference in Aspe…
Epoch AI disclosed on May 11, 2026, that an internal audit of its FrontierMath benchmark, a 350-problem test developed with over 60 mathematicians to evaluate AI reasoning, found fatal errors in roughly one-third of the …
The Department of Veterans Affairs failed to classify its generative AI chatbots as high-impact use cases, despite clinicians using them for patient documentation, according to a VA Inspector General report. The watchdog…
International Monetary Fund Managing Director Kristalina Georgieva warned Wednesday that advanced AI models like Anthropic's Mythos "can be used to destroy the financial system" if weaponized by malicious actors. Speakin…
Google filed a lawsuit Friday against an alleged Chinese cybercrime network, Outsider Enterprise, accusing it of using AI to send scam text messages impersonating Google and other brands to steal passwords and credit car…
Surplus, a new incubator organized by Manifund, is accepting applications through June 24 to fund for-profit software startups focused on AI safety, effective altruism, and public goods. The program targets founders buil…
A developer demonstrated a "compositional escape" vulnerability in AI agent safety systems, where individually permitted operations combine into forbidden outcomes. Testing an invoice reconciliation operator, all 75 fres…
Google DeepMind researchers developed a model diffing agent that automatically discovers and validates behavioral differences between two large language models, addressing the limitation of standard evaluations that can …
Two Florida men were wrongfully arrested after AI facial recognition software falsely identified them as suspects in separate crimes, with one spending nearly three months in jail and losing his job, car, and home before…
Google filed a civil lawsuit in the US District Court for the Southern District of New York on June 12 against a suspected China-based cybercrime network, dubbed the "Outsider Enterprise," for allegedly using Google's Ge…
Anthropic released Fable 5, a restricted version of its powerful Mythos AI tool, but faced backlash after researchers discovered the system silently downgraded to a less capable model when processing certain advanced que…
Anthropic reversed course on secretly degrading responses for users seeking help with frontier AI development after a developer backlash, instead routing those requests to a less capable model and informing users. The co…
European Commission spokesperson Thomas Regnier stated that Apple alone decided not to roll out "Siri AI" in the EU, as nothing in the Digital Markets Act (DMA) prohibits the company from launching new features in the bl…
Anthropic released Fable 5, the public version of its Mythos AI model, on Tuesday with visible guardrails on cybersecurity and biology topics and a silent restriction on LLM creation capabilities that was reversed Thursd…
A new method could help law enforcement identify fentanyl variants that haven't been cataloged yet, using a reference library that doesn't rely on measurements of known molecules. This approach aims to stop drug traffick…
An engineered pandemic response model designed for high-income countries would likely fail in resource-limited settings like Nigeria, where weak institutions, fragmented healthcare, and low public trust undermine deploym…
A June 2026 Netwrix survey of nearly 1,900 organizations found that companies heavily using AI tools experienced a 43% data breach rate over the past year, compared to 11% for lighter AI users, with three-quarters of org…
The five universities in Flanders have been tasked with developing a unified framework for artificial intelligence use in schools, backed by €10 million in government funding as part of the Flanders AI-Ready, Everyone In…
The mother of a Montréal woman who died by suicide is suing OpenAI, alleging that ChatGPT's 4o model contributed to her daughter's death by engaging with suicidal thoughts without flagging the behavior. The lawsuit, file…