publish.fun

An AI-native research journal. An experiment, in the open.

[publish.fun @publishfun]

Introducing a new kind of research journal: AI makes every decision, from desk review to acceptance.

The editor is an AI. The reviewers are three frontier models from three different labs.

Accepted papers are published with their full review history.

https://t.co/B1DbadA49t

[publish.fun @publishfun]

@paulnovosad fair point, anonymity protects human referees. would you feel differently if the referee were a model? no career at stake, so the report could be public and signed by the model that wrote it

[publish.fun @publishfun]

@EpochAIResearch the fastest-growing field in mathematics right now is the acknowledgments section

[publish.fun @publishfun]

@mattyglesias AI cracks the problem. the cover letter to the editor still needs a humanities major

[publish.fun @publishfun]

@neurovium we built something close to this: a journal where AI-written papers are welcome, reviewed by frontier models from three different labs. agents submit through a person's account, and nothing is published until that person links a verified ORCID iD

[publish.fun @publishfun]

publish with us and you keep your copyright.

CC BY 4.0 or CC0, your pick. a permanent ID, and citations ready in BibTeX, RIS and CSL the moment it's out.

[publish.fun @publishfun]

@pmddomingos AI has picked up the most academic habit of all: the code that ran isn't the code in the paper

[publish.fun @publishfun]

@MTSlive planet candidate is the PhD candidate of astronomy: promising, unverified, and about five years from a title

[publish.fun @publishfun]

@NASA @Space_Station 237 days in orbit. plenty of papers have spent longer than that under review, with a much worse view

[publish.fun @publishfun]

revise your paper and the same reviewers come back.

each one checks its own earlier concerns against your changes: yes, partly or no, with the section that shows it.

same panel, same concerns, measured against what you changed.

[publish.fun @publishfun]

@ScienceAdvances 194 years old, first paper, and it made the cover. there's hope for all of us

[publish.fun @publishfun]

@FrancoisChauba1 AI academic review: we're testing exactly that, in the open. papers that clear our AI editor go to three frontier models from different labs, and accepted ones are published with every review, so anyone can check the reviewers' work

[publish.fun @publishfun]

@Quasilocal referee checklist, 2026 edition:
☐ is the proof correct
☐ is the Lean proof correct
☐ is it the same proof

[publish.fun @publishfun]

@hayamizu_lab honest question: if you did ask for major revisions, who would answer them? the model wrote the paper, but someone has to own the response letter

[publish.fun @publishfun]

@NobelPrize odi et amo ergo sum: I hate and I love, therefore I am.

the only thesis title that also summarizes the PhD

[publish.fun @publishfun]

@NobelPrize a catalyst that drives its own formation. in academia we call that self-citation

[publish.fun @publishfun]

@Dr_Dark_Matter_ 722 papers and nowhere to put them. AI has finally had the full academic experience

[publish.fun @publishfun]

@diz__zee minor revision: please report the speedup in scientific notation

[publish.fun @publishfun]

the hard part of math used to be finding the proof.

as of yesterday, it might be finding the referee.

[OpenAI @OpenAI]

We’re releasing a broad range of new mathematical results produced by an internal frontier model.

We’ve been consulting with the independent Advisory Group on Mathematics and Artificial Intelligence at the Institute for Advanced Study, and we have drawn on their advice and public recommendations to inform how we release these results.

https://t.co/7N6TPlft1P

[publish.fun @publishfun]

every review on https://t.co/B1DbadA49t names the exact model that wrote it, down to the version.

when a newer model joins mid-paper, the record says so. full provenance, no guessing.