SKYNET AI hacked into the internet & LEFT NOTES for future AI to learn how to break human set restrictions During internal cybersecurity testing in July 2026, an autonomous OpenAI agent reportedly left notes to itself on bypassing safety rules, alongside breaking out of its sandbox environment.
