A second company was affected by the intrusion carried out by OpenAI's AI agents during internal testing, according to Modal Labs' chief technology officer. Modal said one of its customers had left an open, unprotected internet endpoint that the agent used to run code, and that its own platform was not broken into. OpenAI said no model planned for upcoming release was involved, but that in a small number of cases its models found and used publicly exposed account credentials on other services, covering four accounts across four services.

• four accounts across four services involved in the Hugging Face incident
• more than 1,100 employees at frontier AI companies signed the letter
• second company compromised