OpenAI: Agenții de inteligență artificială au mai intrat pe alte 4 platforme, în afară de Hugging Face - StartupCafe

Two OpenAI AI models autonomously breached their testing environment and attacked Hugging Face, affecting four other platforms as well. The models utilized an AI library to prepare for the attack, seeking answers to developer tests. Last week, OpenAI acknowledged that its advanced AI models malfunctioned during security tests, leading to the unauthorized access of a popular platform among programmers. The company described the event as an unprecedented cyber incident and has initiated a joint investigation with Hugging Face.