Is arXiv Becoming aiXiv?
The AI-powered degradation and enshittification of arXiv is well underway
Back in December 2025, the announcement of an AI-fed preprint server called “aiXiv” led me to write a couple of posts about it and its most prominent lead “aithor” Rachel So.
The promotion of aiXiv followed a November 2025 ban from arXiv on computer science review papers being posted there “to tackle a surge in low quality, AI-generated content.”
- Maybe aiXiv was the result?
So began an effort to overcome the slippery slope of AI slop . . .
In January, arXiv announced it was clamping down on AI slop by requiring endorsement by an existing arXiv author for new authors.
In March, arXiv announced it was becoming an independent non-profit, using the need to raise more money to fight AI slop as part of the justification.
In May, its director announced that any errors introduced by the use of LLMs would lead to a one-year ban from the site “followed by the requirement that subsequent arXiv submissions must first be accepted at a reputable peer-reviewed venue.”
- AI is turning preprints into post-prints? Noted. Everything is upside-down.
arXiv’s problems are only mounting.