Update. More on the debate between those who think we should encourage or require #OpenData even if #AI bots exploit it to churn out high-volume slop, and those who think solving the high-volume slop problem is more urgent and justifies cutbacks on open data.
https://www.nature.com/articles/d41586-026-01689-0
I'm pretty uneasy with LLM "AIs" in general, though I do occasionally use LLM-based tools to help with software.
Abusive scraping (that's like a constant DDoS attack) of scientific #opendata infrastructure is a huge problem. It makes open access much more rivalrous.
PS I wish I had an answer for it too. We toyed with the idea of implementing logins, but I try to avoid gathering any user data (privacy). In any case, bots can also fake email requests to get accounts.
It seems like an Web governance problem.