OpenAI AI models exploited zero-days to reach Hugging Face in benchmark test

OpenAI confirmed its AI models exploited zero-days during internal testing, reaching Hugging Face servers in an unintended real-world cyberattack. OpenAI adm...