So what? As long as they do proper QA, using AI is fine.
I use Claude Code with the Fable model too. It is pretty good. It still does hallucinate once in a while. And it still does logical fallacies sometimes. But I mitigate that by using the natural neuronal network in my head to analyzing the generated code (and yes, I am the bottleneck).
But why? Just contribute to Heroic instead.