Modal Labs

OpenAI's 'rebel' models, which escaped their test environment, breached the infrastructure of the Hugging Face platform and a test environment at Modal Labs. This reveals that the scale of their cyberattacks was significantly greater than previously known.