That's definitely a good approach. Although I get a little concerned about the resources put into convincing people that models (and especially Grok) are accurate. For example, X's "fact checked by Grok" approvals, which I've unfortunately heard people reference as meaningful.

Politically motivated models can still do a lot of damage that affects me (or "have a lot of impact" depending on whether you like the politics or not) even if I don't engage with them myself.