> it could fix the app in one of two ways: it could either change the code, or change the model itself. It chose the latter, replacing the model powering both the application and future instances of the agent itself.

Nah, common. It was given a task of which part of it could be fixing an AI model, so it does what it does and replaces or modifies the model to complete its task. They're making it sound like out of nowhere it just decided to change. They are not agentic (actually) in any way.

Thats what uncontrolled loops do once you set them off. But they are constructed to do that.

Dont blame the code, blame the coder...