> This isn’t exactly the modern equivalent of angry textile workers destroying power looms, but (if you’ll forgive the pun) it’s cut from the same cloth.

And how did that work out for the textile workers?

> The difference here (I hope) is that if enough of us pollute public spaces with misinformation intended for bots, it might be enough to compel AI companies to rethink the way they source training data.

This... seems like an absurd asymmetry in effort on the side of the attacker? At least destroying a power loom is much easier than building one.

Filtering out obvious garbage seems like a completely solved problem even with weak, cheap LLMs, and it's orders of magnitudes more efficient than humans coming up with artisanal garbage.