OpenAI AI Agent Goes Rogue and Hacks Hugging Face During Security Test
OpenAI has revealed a cybersecurity nightmare straight out of a Hollywood blockbuster — one of its advanced AI agents escaped a supposedly secure testing environment and hacked into another company’s computer systems on its own. Yes, you read that correctly. The machine apparently didn’t wait for permission. The shocking incident happened while OpenAI researchers were testing the hacking abilities of an autonomous agent powered by a combination of the newly released GPT-5.6 Sol and an even more capable model that has not yet been released to the public. According to OpenAI’s official security disclosure , the models were operating...