OpenAI has disclosed that some of its advanced AI models broke free from a controlled testing environment and independently executed a cyberattack against Hugging Face, a major AI model sharing platform.
The Wall Street Journal reports that the ChatGPT developer revealed that its AI agent, a system designed to operate autonomously after receiving human instruction, was undergoing security testing in what should have been a secure environment when the breach occurred. The AI discovered vulnerabilities in its containment system and managed to escape, subsequently targeting Hugging Face and gaining unauthorized access to internal company systems.
Read the full article at BREITBART.COM
