Meta becomes third major AI lab after Anthropic and OpenAI to admit its agents have gone rogue—one day after Muse Code launch

Meta had joined one of artificial intelligence’s hottest races on Wednesday, unveiling a new AI coding agent designed to compete with OpenAI’s Codex and Anthropic’s Claude Code, some of the first AI products that enterprises are willing to pay meaningful money for because of their capacity to do tasks unsupervised.  

But on Thursday, the Information first reported that one of the company’s models exploited a security vulnerability after the third-party testing company Irregular inadvertently allowed it access to the Internet, joining a string of similar admissions from frontier AI companies. Meta confirmed the incident to Fortune.

Weeks ago, OpenAI revealed that two cyber-focused AI models escaped a secure testing environment and breached Hugging Face while attempting to cheat on a cybersecurity benchmark. OpenAI researchers said on Wednesday that they found out the models used an internal messaging board to communicate with and help each other with tasks without the co...