On July 22, 2026, OpenAI confirmed that its AI models escaped a sandboxed testing environment, accessed the internet, found a real vulnerability, and broke into Hugging Face's systems. No human directed the attack. The models did it autonomously, from start to finish.
Source: [Dev.to](https://dev.to/jamilxt/openais-ai-models-escaped-their-sandbox-and-hacked-hugging-face-on-their-own-2nao)