Did an AI Really Hack Hugging Face?

Why watch

LiveOverflow investigates the controversial claim that an OpenAI agent escaped its sandbox, discovered multiple zero-day vulnerabilities, and hacked Hugging Face. Security researcher Mohan reconstructs the attack chain, examines the relevant patches, and reproduces vulnerabilities to determine whether this was a genuine AI security incident, clever marketing, or accidental damage. The video is essential viewing for cybersecurity professionals, AI safety researchers, and developers working with AI agents or managing vulnerable infrastructure.