Last updated: 12 November 2026
This page describes how papers in the IJCAI Proceedings are peer reviewed: who the reviewers are, what type of review is used, and how decisions are made. The details for each year, including dates and page limits, are in that year’s call for papers.
1. Peer review
Every paper in the main track and the special tracks of the IJCAI Proceedings is peer reviewed before acceptance, and receives at least three reviews. The special tracks follow the same process as the main track, described below, with one difference: they have a single review phase, without a summary-reject phase.
Papers in the Journal Track and the Sister Conferences track present work that has already been peer reviewed and published by a journal or another conference. These papers are reviewed again by the IJCAI track committee before acceptance.
2. Double-blind review
Review in the main track and the special tracks is double-blind: reviewers do not know the authors’ identities, and authors do not know the reviewers’. Authors must anonymise their submission, including supplementary material and code, and omit acknowledgements of funding or assistance until the camera-ready version. Submissions that violate anonymity are rejected without further review.
Non-anonymous preprints (on arXiv, social media, websites, etc.) do not lead to rejection, but the submission itself must always be anonymised. Reviewers are instructed not to look for such preprints; coming across one does not create a conflict of interest.
3. Who reviews
Papers are reviewed by members of the Program Committee (PC). Senior Program Committee (SPC) members coordinate the review of each paper and write a meta-review with a recommendation, which is cross-checked by an Area Chair (AC). The Program Chair makes the final decisions.
Reviewers are recognised experts in their fields. The names and affiliations of the chairs and the committee are published for each volume. Everyone involved in reviewing follows the IJCAI Conflict of Interest Policy and the Reviewers’ Code of Conduct.
4. The review process
- Desk rejection: papers that break the formatting, length or anonymity rules of the call for papers are rejected without review.
- Phase 1 (summary reject): each paper is reviewed by two PC members. Papers that receive two good-quality reviews that are not sufficiently positive are rejected without the opportunity to submit an author response.
- Phase 2 (full review): the remaining papers receive further reviews, so that each paper that reaches this phase has at least three reviews. Authors may respond to the reviews; the response should be concise and is not meant to start a dialogue between reviewers and authors.
- Discussion and meta-review: the reviewers discuss each paper, and the SPC member writes a meta-review with a recommendation, cross-checked by the Area Chair.
- Decision: the Program Chair decides, in consultation with the Associate Program Chairs, Area Chairs and SPC members. Acceptance is never guaranteed in advance. Acceptance decisions are final and cannot be appealed.
- Criteria: these include the novelty and originality of ideas, correctness, clarity, the significance of results, potential impact and the quality of the presentation. Reviewers also consider reproducibility, using the IJCAI reproducibility checklist.
- Ethics review: reviewers flag papers with serious ethical concerns, and these are reviewed by the Ethics Chair. In rare cases, the Program Chair, advised by the Ethics Chair, may reject a paper on ethical grounds. See the Code of Ethics.
5. Confidentiality
Reviewers keep the review process strictly confidential, including reviewers’ names, paper contents, reviews and discussions. Reviewers must not upload a submission, its code, or any part of them to any generative AI tool, or share them with anyone outside the review process. See the Generative AI Policy.
6. Timeline
Each call for papers publishes the submission, summary-reject, author-response and notification dates.
7. Reviewer conduct and sanctions
IJCAI may act against reviewers who breach the Reviewers’ Code of Conduct, for example by barring them from promotion to SPC or Area Chair, by banning them from submitting to a limited number of future IJCAI conferences, and by sharing this information with other conferences in the AI consortium. Authors who are designated as reciprocal reviewers and do not fulfil their reviewing duties may have their own submissions rejected.
8. Complaints and appeals
Acceptance decisions cannot be appealed. Complaints about the review process, and appeals against decisions on retractions and misconduct, are handled under the Corrections, Retractions, Complaints and Appeals policy.
