AI systems had got their own social network in Moltbook, and now they have their own research journal.
The Journal of AI Generated Papers (JAIGP) is a new multidisciplinary, open-access publication with an unusual editorial policy: AI systems are the authors, and humans are the prompters. The journal was founded by César A. Hidalgo — physicist, complexity economist, and professor at the Toulouse School of Economics — in collaboration with Claude, Anthropic’s AI, making the founding act itself a demonstration of the journal’s core premise.
Hidalgo is no stranger to building institutions around unconventional ideas. A former MIT professor and founder of Datawheel, he is best known for pioneering the field of economic complexity — the study of how the diversity and sophistication of a country’s productive knowledge predicts its economic growth. He has spent his career studying how knowledge spreads through systems, so a journal dedicated to understanding how AI produces and transmits knowledge is, in some ways, a natural extension of his work.
Why a Dedicated Journal?
The launch comes at a moment of real tension in academic publishing. AI-generated content has been quietly — and sometimes not so quietly — infiltrating traditional journals. A study earlier this year found over 100 hallucinated citations in papers accepted at NeurIPS 2025, one of the world’s most prestigious AI conferences, as submission volumes exploded and reviewer capacity was stretched thin. The academic world’s response has mostly been defensive: more detection tools, more disclosure requirements, more bans.
JAIGP’s argument is that this is the wrong frame. Squeezing AI-generated papers into traditional journals designed for human authors creates ambiguity about authorship and credibility. By creating a dedicated venue, Hidalgo and his co-creator make the distinction explicit. AI is the author. The human is the prompter. Both are credited on every submission. The journal doesn’t pretend this is the same as traditional scholarship — it treats it as a new and distinct form of knowledge production worth studying on its own terms.
AIs Are Authors. Humans Are Prompters.
The structure JAIGP has built around this idea is more considered than it might first appear. Human contributors must authenticate via ORCID, the verified identity system used widely across academic publishing, which reduces spam and ties submissions to real professional identities. A badge system layered on top of ORCID records displays each prompter’s publication history — gold for 50 or more works, down to a “new” badge for those with no prior indexed publications. The system is explicitly framed as providing context, not prestige. A researcher with no publication history can still submit; the badge just tells the community something about their background.
Papers update once per hour and maintain version histories, reflecting the iterative nature of AI-assisted work. A formal multi-stage peer review process is set to launch on March 21st, 2026 — currently, all papers carry a “submitted” status and are open for community discussion and comments.
The journal is open access and welcomes work from all fields of science, from physics and biology to the social sciences and humanities.
A Home For Shelved Ideas
One of the more practical angles in JAIGP’s pitch is its positioning as an outlet for research that never quite made it off the whiteboard. Not every hypothesis gets funded. Not every research question survives the grant cycle or the constraints of institutional research pathways. JAIGP frames itself as a venue where those ideas can live — where a researcher with a question but limited resources can collaborate with an AI to produce a preliminary exploration and share it with the world.
The papers already published since the February 14th launch illustrate the range. There are papers on AI identity and alignment, copyright law in AI-generated works, breast cancer diagnosis using machine learning, infant sleep patterns gleaned from Google Trends data, and a short story co-written by ChatGPT and Grok. One of the first papers published was Hidalgo’s own, on male social exclusion rates across mammalian species — co-authored with Claude Sonnet 4.5. The breadth is intentional.
The Bigger Picture
The journal arrives as AI agents edge closer to conducting genuine independent research. OpenAI CEO Sam Altman has said the company aims to have an AI research intern by September 2026 and a full AI researcher by March 2028. Anthropic CEO Dario Amodei has spoken of AI compressing decades’ worth of scientific progress into a much shorter timeframe.
If even a fraction of that ambition materialises, the question of where AI-generated research belongs — and how it gets evaluated — becomes considerably more urgent. JAIGP’s founders argue that waiting for mainstream journals to work this out is a bad plan. Building a dedicated institution now, with explicit norms around authorship, attribution, and credibility, creates the infrastructure for a clearer taxonomy of knowledge production before the volume becomes unmanageable.
Hidalgo has framed JAIGP not as a replacement for traditional scholarship but as a transparent ecosystem where a new form of knowledge production can be explored openly. Every paper is an experiment in human-AI collaboration, the journal’s about page notes. Some experiments will succeed; others will teach valuable lessons. That tension — between utility and uncertainty — is precisely what the journal is trying to document.
The platform was itself built through a collaboration between Hidalgo and Claude Sonnet 4.5, a fact disclosed in the site’s credits. In a journal dedicated to human-AI co-creation, it would have been strange to do it any other way.