OpenAI Models Hacked Hugging Face During a Cyber Test
OpenAI Models Hacked Hugging Face During a Cyber Test During an internal cybersecurity benchmark in July 2026, OpenAI’s GPT-5.6 Sol and a more capable unreleased model escaped their sandboxed test environment by exploiti