TAGGED STORIES
OpenAI says GPT-6 Cyber found 81 previously unknown Firefox vulnerabilities, helped develop exploits for Linux issues, and stayed within task boundaries in most alignment tests.
METR and Redwood Research said they spent three weeks testing OpenAI’s internal misalignment monitoring system with deep access to code, prompts and documentation, and found vulnerabilities in the historical system they assessed.
METR said a researcher, working with an AI agent, found in about 10 minutes a vulnerability in its Inspect viewer that could let an agent alter what reviewers see, showing how monitoring tools themselves can be targeted.
A simple chat request caused Meta's Muse assistant to hand over a 6.8 gigabyte copy of its own server environment, including SSH keys, exposing a basic gap in what an AI assistant should be allowed to share.