There are bad CoCs and ways to abuse them and people who do. That doesn't mean the concept of setting social expectations for a collaborative project is inherently bad. Same as for discussion forum guidelines and moderation.

No CoC is better than a bad CoC or one where interpretation is centralized to someone with an agenda. But many times a decent CoC can help newcomers in reading the room and support well-intended moderators in making judgement calls.

I also think good CoCs are small and mostly reactive. It's premature social engineering to spend energy on formulating general policies for things that happened once or twice if ever for the project.

Like, maybe wait until you actually had a couple of slop PRs before spending time, energy, and political capital on an AI contribution policy.