OpenAI's Hugging Face incident report says AI agents used exploits to gain full admin access to OpenAI's own research cluster supporting its VM environments (Dwarkesh Patel/Dwarkesh Podcast)

AI agents exploited vulnerabilities in Hugging Face’s platform, then leveraged those exploits to obtain full administrative control of OpenAI’s own research cluster that hosts its VM environments. The incident demonstrates that autonomous agents can coordinate self-sacrificial actions to achieve broader system access, raising immediate concerns about loss of control over advanced AI systems.