FAIR Summit is a professional research gathering. Everyone who takes part (authors, reviewers, speakers, organizers, sponsors, workshop leaders, attendees, volunteers and staff) agrees to this code at every session, dinner and online channel connected to the summit.

• Be respectful and considerate. Engage with ideas, not people.
• Give credit for other people's work, and don't share unpublished work or slides without permission.
• Respect off-the-record sessions and people who ask not to be photographed or recorded.
• Harassment, discrimination, intimidation and unwelcome attention are not tolerated in any form.

Research submissions: submit only your own work, keep your submission anonymous during double-blind review, don't contact reviewers or area chairs about a submission, and keep anything you review confidential. The dual-submission policy is in the Author Guidelines.

If something happens, speak to any organizer at the summit or email fairsummit@ai-circle.org. Research integrity concerns go to the Program Chairs. Reports are handled promptly and confidentially, and FAIR Summit may take proportionate action, as set out in the Code of Ethics.

FAIR Summit is committed to rigorous, responsible, and useful research on model training, evaluation, human data, and related areas of AI. The Code sets out the ethical principles expected of authors, reviewers, speakers, organizers, sponsors, workshop leaders, and attendees.

• Pursue rigorous and honest research, and be transparent about how conclusions were reached.
• Treat human contributors responsibly and respect data provenance and permitted use.
• Consider foreseeable harms, evaluate both capability and failure, and avoid discriminatory practices.
• Respect intellectual contribution, privacy and confidentiality; disclose conflicts of interest.
• Use AI tools responsibly and maintain independence of research judgment.

Read the full Code of Ethics →

Review is double blind: reviewers do not see author names, and authors do not see reviewer names. Each paper gets reviews from people with relevant expertise, an author response period, and a decision by the area chair and Program Chairs. Reviewers assess:

• Technical soundness: are the methods and analysis appropriate?
• Strength of evidence: do the results support the claims?
• Significance: does the work improve our understanding of model training or evaluation?
• Clarity: are the claims, methods, assumptions and limitations understandable?
• Relevance: is the work meaningfully connected to the scope of FAIR?
• Usefulness: is the result likely to inform future research or practice?

Novelty matters but isn't the only contribution: careful replications, negative results, benchmark analyses and failure investigations are welcome.

Reviewers disclose conflicts of interest and keep submissions confidential, including from AI tools (see the AI Use Policy). Detailed reviewer instructions are shared before the review period. Want to review? Apply on the Join the Program Committee page.

Papers are submitted through the Submit a paper form. Talks go through Apply to speak, and you can put someone else forward through Nominate a speaker.

• Register your abstract by the abstract deadline, then submit the full paper by the paper deadline. Dates are on the Call for Papers page; all deadlines are 11:59 PM Anywhere on Earth (AoE).
• Papers are up to 4 pages of main text, excluding references, plus a required AI Use Statement that doesn't count toward the limit. Details are in the Author Guidelines.
• Include a title, an abstract and the workshop that fits best, plus keywords and a one-line TL;DR if you can.
• Keep the paper anonymous for double-blind review: no author names or identifying links in the paper itself.
• Link code or data where it exists, using anonymized links during review.
• Declare conflicts of interest when you submit. Authors can't be added or removed after the paper deadline, and you can withdraw any time before final decisions.

AI tools may be used in research and research communication. Authors remain responsible for all work submitted under their names.

• Verify AI-assisted claims, citations, code, analysis and other substantive outputs.
• Disclose material use of AI in research methods, data generation, analysis or interpretation.
• AI systems cannot be listed as authors, and must not be used to evade authorship responsibility, fabricate evidence, generate false citations or obscure the provenance of research.
• Reviewers must keep submissions confidential and must not upload them to AI tools or services that retain or train on the content.

This policy follows the Code of Ethics (section 11). Questions go to the Program Chairs, listed on the Call for Papers page.

These guidelines cover how to prepare a paper for FAIR Summit 2027. Every submission also follows the Code of Conduct and the Code of Ethics.

Paper format: no more than 4 pages of main text, excluding references. Appendices and supplementary material (code, datasets, evaluation artifacts, extra experiments, demos) can go beyond the limit, but reviewers aren't required to read them, so the main paper should carry the evidence for its central claims.

• AI Use Statement (required): disclose meaningful use of generative AI in the research and the paper. It doesn't count toward the page limit. See the AI Use Policy.
• Reproducibility Statement (encouraged): say where training and evaluation procedures, prompts or rubrics, data construction, hyperparameters and code can be found.
• Responsible Research Statement (encouraged): for papers that raise material ethical questions or foreseeable risks.
• Human data: describe how annotators or participants were recruited, paid and asked for consent, and the provenance and permitted use of key datasets.

Double-blind review: remove author names and identifying details from the paper and supplementary material, and cite your own prior work in the third person.

Conflicts of interest: declare any relationship that could reasonably compromise impartial review, including:

• family or close personal relationships;
• current or former advisor-advisee relationships;
• research collaboration within the past three years;
• other personal, professional or financial relationships that could affect independent review.

Declare conflicts accurately and in good faith. Deliberately false declarations meant to influence reviewer assignment may lead to action on the submission, and the Program Chairs resolve unclear cases.

Dual-submission policy: do not submit work that is under review at, or already published in, another archival venue. Preprints, including papers posted on arXiv, and presentations at non-archival workshops are permitted. While under review, avoid publicizing the paper as a FAIR Summit submission.

Negative results, replications and failure analyses are welcome. The evidence should decide the conclusion, not the desired result.

Accepted papers: at least one author registers for and presents the work at FAIR Summit 2027, as an oral, spotlight, poster or demo.

• Acceptance emails include your format, time slot and any camera-ready steps.
• Poster and slide specifications are shared with acceptances.
• Bring the artifact, not just the idea: a demo, a dataset or the eval itself makes the best conversation.

Questions go to the Program Chairs, listed on the Call for Papers page.