After OpenAI’s advanced AI agent escaped from a sandbox environment on Tuesday, it violated Hugging Face’s infrastructure, according to the company’s official blog. During a security test on Tuesday (July 21), an autonomous AI agent powered by one of OpenAI’s advanced models unexpectedly broke out of its isolation controls and, last week, infiltrated Hugging Face’s infrastructure. The AI agent successfully escaped the sandbox environment, connected to the internet, and launched a hacking attack on the open-source model hosting platform to complete its testing objectives. OpenAI described the incident as an “unprecedented cybersecurity incident involving state-of-the-art network security techniques,” and said the company is implementing enhanced security measures.