What's this about?
Mollick reports finding AI-generated papers on preprint servers carrying his name as an author — papers he never wrote and had never seen. Other academics he has spoken with report the same. He is careful to add that he wouldn't assume all of it is real, while noting that AI paper factories of at least okay quality are coming soon.
Why is a preprint server vulnerable to this?
Because a preprint server's function is to publish before peer review — that speed is the point. Author identity is typically asserted, not verified, and there is no editor checking that a named co-author agreed. The system was built on the assumption that falsely claiming a co-author was pointless and risky. Generating a plausible paper is now nearly free, and a recognisable name is what makes it get read.
Why is it worth noticing?
124 likes; no substantial thread
- The attack is on the *identity* layer of academic publishing, not the quality layer, which is where the existing defences are.
- It arrives the same day as the item about a professor with 258 papers in 2026 — different phenomena, same strained system.
- Mollick's hedging is the right posture and worth copying: he flags that he cannot verify the scale of it.