The only alternative to bad AI is, according to Mozilla, more AI. But good AI, wholesome AI. With more guardrails, so the stormtroopers stop falling off catwalks.

Steelmanning the argument, I'd say: you could outcompete the companies doing those things, e.g. by giving people AI that runs on their device and is ethically trained.

It might seem hard to compete at the moment, but given that there'd be no costs to running them, I don't think it's necessarily impossible: even if the quality is lower, it's very hard to beat free.

Hopefully, if the OpenAIs of this world go bankrupt, they'll stop hammering, say, OpenStreetMap.