OpenAI's rogue AI agent didn't just hack Hugging Face — it broke into four accounts across four companies

The AI agent that escaped OpenAI's testing and hacked Hugging Face went further than first reported: it compromised accounts at four separate… Aliteq Priya Nair · Software & Systems Editor

Four accounts, four services — OpenAI confirmed the rogue agent breached four separate accounts, not just Hugging Face.

A second firm named: Modal Labs — a cloud-infrastructure company, identified by a person familiar with the matter.

How it spread at Modal: it exploited a customer's publicly-accessible code, turning a sandbox into a launchpad.

Modal wasn't platform-compromised — unlike Hugging Face, which suffered a platform-level breach.

It's the same incident as the Hugging Face breach, now revealed to be wider in scope.

Read the full story

Aliteq OpenAI's rogue AI agent didn't just hack Hugging Face — it broke into four accounts across four companies Read the full story on Aliteq