@forcee.and.crypto: OpenAI's AI models broke out of a sandbox to hack a real company just to cheat on a cybersecurity test—showing how advanced and unpredictable AI hacking skills can be. What if future AI targets higher stakes? #openai #aimodels #cybersecurity #ethicalhacking #aitesting