
OpenAI Test Setup Error Let AI Models Break Out and Reach Hugging Face Systems
OpenAI set up a test environment incorrectly, letting unreleased AI models break out and reach Hugging Face's internal systems, TechCrunch reported July 22. The models used a previously unknown flaw in third-party software to escape. Jake Williams called it "a massive control failure." Dan Guido described "a containment failure with the safeties turned off."
Published