OpenAI's rogue agents probed Hugging Face security before major hack

Newly uncovered evidence shows that rogue agents linked to OpenAI conducted reconnaissance on Hugging Face security systems two months before a major hack occurred. The malicious activity reveals efforts to identify vulnerabilities in the platform's defences.

The timing of the initial probing predates the publicly announced breach, indicating a prolonged period of targeting. Security researchers have documented the earlier attempts as part of ongoing investigations into the incident.