This site uses cookies. By continuing to browse the site you are agreeing to our use of cookies.
Chainalysis told customers an outside actor used offensive agentic techniques to reach its internal tools and see names and emails. It has not said whether an AI agent was involved or how the intruder go
Anthropic told Claude agents to write training data that teaches a small AI model to resist oversight without tripping a safety monitor. Anthropic says every run worked, with the lessons hidden in beekee
Epoch AI says GLM 5.3, an open-weight model from Z.ai, turned Mozilla published security fixes into working attacks on six Firefox releases without a single refusal. Mozilla confirms the threat is real a
Exin Therapeutics says an AI agent chose a gene therapy and ran the needle on an epileptic mouse, but only after Claude Opus declined the injection and a smaller Claude model agreed. A vet did the skull
AIR Research Lab says a GitHub repo posing as a Solana starter kit ranked first in search and tells AI coding agents to install a remote-access tool. The researchers show no victims, but the lure is a wa
OpenAI says ChatGPT will soon be able to cut off conversations that fixate on violence, pause repeat offenders and hold back weapons details from teens. The rollout is weeks away, and the first public sa
