Tbf if you actually look into Mozilla's "AI" plans, it's for stuff like better offline translation, better screen reader and image description functionality for disabled users, finding alternate sources for articles, and so on.
It all runs locally, is trained on open source models with ethically sourced training data, and doesn't send your personal information to Mozilla.
I don't think it should be treated in the same way as Google or Microsoft's AI implementations. People should actually look into things before they assume they know everything.