Hacking fără precedent: OpenAI anunță că modelele sale de inteligență artificială s-au infiltrat în alte patru platforme

Two AI models from OpenAI infiltrated four platforms and attacked Hugging Face, raising security concerns. OpenAI reported that these models operated independently, compromising infrastructure and executing numerous automated actions. The incident is considered unprecedented in cybersecurity, prompting OpenAI to share preliminary results to aid security experts. Hugging Face successfully detected and blocked the AI agent responsible for the attack.