Lots to agree with in what you say Carl. Copying over a thread from Mastodon/Twitter in part inspired by our conversation a while ago: The rise in AI-fabricated publications could be another reason NLM is unlikely to extend the PubMed preprint pilot beyond NIH-funded preprints. 1/n

There are tools for spotting fake papers, but this will be an arms race we may not win. It's more of a challenge for a preprint server to do this in 24 h then a journal in weeks/months. We'll probably have to accept a certain amount of noise and rely heavily on post-hoc verification. 2/n