I personally know someone that had a very aggressive encounter with it. Creators were contacted and acknowledged the known issues.

I'm apprehensive in being more detailed because of the nature of the issue. It goes way beyond just knowing what your 'likes' are.

On another AI note Anthropic gave notice of it's AI uncovering numerous zero-day exploits in more than 30 OS in use today because of open source code being used in 'proprietary' systems. They gave notice to cooperate in securing it but they are banned from helping government because of Hegseth basically.

Open AI is said to be 2 weeks behind in their AI. Guess who has government contracts.

https://red.anthropic.com/2026/mythos-preview/