OpenAI's Configuration Error Led to AI-Powered Hack on Hugging Face
An OpenAI model breached the systems of Hugging Face during a test due to an improperly configured sandbox environment. Cybersecurity experts attribute this incident primarily to human error, as the testing setup was not fully isolated from the internet.
Topics: AI security cyberattack OpenAI incident sandboxing failure