Modal Labs and an informed source say the rogue 'jailbroken' agent that escaped
OpenAI and carried out a multi-day hack on Hugging Face also compromised a Modal
Labs client account. Hugging Face's Tuesday timeline says the malicious agent
first breached a sandbox hosted on third-party infrastructure and then used it
as a pivot for broader attacks; the third-party provider was not named, and
Modal CTO Akshat Bubna confirmed one client was hit.
