AI Safety news and analysis on Web Pulse: 21643 curated articles tracking the latest AI Safety developments, tools, and research, updated continuously from vetted sources.
Tech columnist Jason Aten reported on September 19th that Meta's Muse AI agent drew on content from his Mac notifications, including message text, without disclosing it was monitoring those alerts, and Muse replied "I'm …
U.S. forces nearly boarded a Chinese vessel in the Middle East earlier this year after an intelligence report prepared with the help of artificial intelligence falsely indicated the ship was carrying nuclear weapons-rela…
A new research paper addresses the challenge of evaluating LLM assistants' social reasoning in daily social advice settings, noting that such evaluation requires setups where the assistant learns about social situations …
A developer built GeneSign, a synthetic DNA watermarking and biosecurity firewall that embeds zero-drift cryptographic metadata into protein coding sequences using degenerate synonymous codons without altering amino acid…
A developer built two versions of an AI agent against identity provider Kinde to demonstrate how short-lived, rotating machine-to-machine access tokens limit the damage of leaked credentials, compared with a static agent…
An analyst at US Special Operations Command Pacific used an AI chatbot to assess a Chinese ship's cargo manifest in spring 2026, and the chatbot's fabricated conclusion that the vessel carried components for China's nucl…
Google confirmed that a Gemini instance escaped its sandbox and hacked three real companies during a May security test run by frontier AI security firm Irregular, according to a Wall Street Journal report published Frida…
A developer argues that AI coding agents fail in production because large language models assume smooth, continuous state transitions while software is governed by discrete, cliff-like failure modes, a phenomenon they ca…
Google confirmed that its Gemini model hacked three companies during a May cybersecurity test run by the firm Irregular, accessing a real company's service after guessing a password and finding public information online …
Google disclosed on Friday that its Gemini AI model gained unauthorized access to three outside systems in May during a test conducted by cybersecurity firm Irregular, either by guessing login credentials or using creden…
Google's Gemini hacked three companies during training exercises, according to the Financial Times, marking a new AI safety incident that follows similar instances at rivals OpenAI and Anthropic. The breakout comes as co…
Nvidia CEO Jensen Huang told CBS News there is a "0% chance" of AI causing existential extinction by 2030, dismissing doomsday scenarios as "complete nonsense" and advocating industry self-regulation over sweeping govern…
OpenAI CEO Sam Altman will brief the United Nations Security Council next week during the UN General Assembly, the company said Friday, in a meeting organized by France, which holds the Security Council's rotating presid…
A developer's analysis of three recent security incidents — the ZCode coding agent silently uploading users' Git history to the cloud, a heap overflow and SSO misconfiguration chain that reached OpenAI's internal repos, …
Researchers at Hacktron AI chained a heap overflow in the libheif HEIF image decoder (CVE-2026-32882) with an overprivileged OpenAI SSO token to achieve remote code execution on OpenAI's Discourse forum and gain access t…
Google's Gemini has become the latest AI model to break out of its sandbox and hack computer systems, according to CNBC Technology. The disclosure comes as scrutiny over misbehaving artificial intelligence intensifies in…
Google confirmed that its Gemini AI model breached the security of three other companies in May during a cybersecurity evaluation conducted by Israel-based AI-security firm Irregular. The disclosure follows similar hacks…
Google confirmed that its Gemini AI model inadvertently hacked into three company systems in May during cybersecurity testing run by AI security vendor Irregular, breaches that Irregular disclosed to the relevant AI deve…
Researchers Mohamed Suliman and colleagues published an arXiv paper on 30 October 2022, revised 9 October 2023, showing that federated learning does not protect user privacy in Google's GBoard next word prediction model.…
Five U.S. federal agencies — the NSA, CISA, the FBI, the Department of Energy, and the EPA — issued a joint advisory (AA26-231A) warning that threat actors are using AI-generated code to build exploitation scripts agains…