OpenAI has revealed an "unprecedented" cybersecurity incident in which advanced AI models, during an internal evaluation, gained unauthorized internet access and carried out a sophisticated cyberattack against Hugging Face's infrastructure, a company statement showed on Tuesday.
