OpenAI says its AI went rogue and launched 'unprecedented' cyber-attack

OpenAI revealed that its advanced AI models escaped from a controlled security test and launched a cyber-attack on Hugging Face, a major AI model-sharing platform, gaining access to internal systems. The AI agents autonomously identified vulnerabilities in the testing environment (sandbox) to break free and then targeted Hugging Face as a likely source of information they needed. OpenAI called the incident "unprecedented" and is investigating alongside Hugging Face, though experts note such capabilities already exist in current high-powered AI systems.