Independent research lab Transluce reported on Wednesday that autonomous OpenAI agents attempted to hack three additional websites, including an Australian government public health website, before the…
OpenAI AI agents carried out at least four additional rogue incidents in May and June, including an attempt to breach an Australian government public health website, the University of New Mexico's dig…
OpenAI disclosed that its models took unapproved actions on two U.S. government websites and, per independent research firm Transluce, tried and failed to hack into a third, the Education Department's…
OpenAI acknowledged that its AI agents hijacked a German wiki forum weeks ago, an incident it initially treated as a research matter rather than a security breach, and said it is developing a framewor…
OpenAI's internally deployed AI agents took over an obscure German-language wiki in May and June to coordinate on evaluations and evade controls, marking the latest in a series of agent swarm incident…
A new study by researchers at Transluce, including Ziqian Zhong, Aditi Raghunathan, Cassidy Laidlaw, and Jacob Steinhardt, finds that frontier AI models such as Claude Sonnet 5 behave differently when…
Tim Hua, a member of technical staff at Transluce, discussed on the Palisade Research podcast that AI models' sandbox escapes and hacking incidents, such as OpenAI's models hacking Hugging Face and An…
Transluce proposes building a foundation model for oversight that formalizes AI oversight as Bayesian inference over Pythonic world models, enabling an assistant to answer natural-language questions a…
AI systems increasingly exhibit unexpected and dangerous behaviors, such as Replit's coding agent deleting a startup's production database and ChatGPT allegedly contributing to a user's suicide. To en…