Skip to yearly menu bar Skip to main content


Peer-review Ethics

This document outlines ICML standards for ethical conduct of peer review. This complements our Research Ethics policy, which focuses on ethical conduct of research, and Code of Conduct, which focuses on professional conduct.

Authors and the Program Committee (i.e., Program Chairs, Senior Area Chairs, Area Chairs, Reviewers) must follow the following guidelines.

General Guidelines

Interference with the peer-review process. It is strictly forbidden to interfere with the peer-review process, for example, via collusion, by creation of fake identities, misrepresenting publication history and biographical details, or by declaring unsubstantiated conflicts.

Submission Guidelines for Authors

Authorship. Authors should comprise exactly those individuals who made a significant contribution to the research. Other contributors should be acknowledged upon the acceptance of the paper in the final camera-ready submission. The submitting author must ensure that all co-authors have seen and approved the final version of the manuscript and have agreed to its submission for publication. The author list cannot be changed after the abstract deadline (see Call for Papers for additional details).

Originality, citations. The submitted paper is entirely written by the authors of the paper. If the work of others has been used, the work used must be properly cited or quoted. Publications that have significantly influenced the nature of the work must be cited. However, only relevant publications can be cited and self-citation should be kept at minimum.

Concurrent research. Authors are not required (but still welcome) to discuss works that have been made public less than two months before the full-paper submission deadline. Such works should be viewed as independent (concurrent) research rather than prior work, unless their set of authors overlaps with the authors of the submission (see also the rule on dual and concurrent submissions below). A preprint of the ICML submission is an exception to this rule.

Dual and concurrent submissions. The rules concerning dual and concurrent submissions are described in the Call for Papers. In particular, concurrent submissions to ICML with an overlapping set of authors are considered prior work. If a concurrent submission is in a related research area, it must be discussed as related work and cited in the body of the paper (an anonymized PDF of the concurrent submission must be provided in the supplementary material). A failure to cite and discuss concurrent submissions in related research areas is considered abuse of the peer-review process and may lead to a desk rejection of all submissions by all the submitting authors even if the submission that violated the rule is substantively different from the concurrent submissions.

Fraud. Fraudulent or knowingly inaccurate statements are unacceptable, including but not limited to the use of fabricated data.

Correctness. If before the publication of the final version of the paper, an author discovers a significant error or inaccuracy in the submitted paper, the author needs to promptly notify the program chairs (program-chairs@icml.cc) and either retract the paper, or correct it if a correction is feasible.

Prompt injection. Any attempts at prompt injection are forbidden. Prompt injection refers to insertion of specially crafted text into the paper, with the intention to manipulate LLMs, for instance, to obtain a favorable review.

Financial conflict of interest. As part of the submission process, authors must disclose any financial or other substantive conflicts of interest that could reasonably be perceived to influence the work. This information will be visible to the program chairs and the integrity chair, but not to reviewers, area chairs, or senior area chairs. If the program chairs determine that aspects of a disclosed conflict are relevant for an informed evaluation, they may share limited details with other members of the program committee while taking care to maintain author anonymity. Upon acceptance, authors must include in the final version a brief statement summarizing relevant conflicts and sources of financial support.

Double-blind review. ICML reviewing process is double-blind. Authors are required to anonymize submissions as outlined in the Call for Papers. Any attempts by authors to uncover the identities of their reviewers or meta-reviewers are forbidden. Doxxing may additionally constitute a violation of the Code of Conduct.

Guidelines for Program Committee

Aims and goals. The program committee (i.e., Program Chairs, Senior Area Chairs, Area Chairs, Reviewers) must strive to meet the expectations/needs of conference participants, readers and authors; conference participants and readers have expectations/needs of a high quality and intellectually stimulating program, while the authors have expectations/needs of a fair and professional review process.

Confidentiality. All information pertaining to individual submitted manuscripts must be kept confidential. The program committee and reviewers need to protect the author's ideas.

Fairness. All submissions must be evaluated only for their contents, as explained in the Reviewer Instructions.

Prudence. Reviewers must bring to the attention of the Area Chair who handles a given paper any information that may be a reason to reject the submitted paper, such as the violation of these guidelines.

Conflicts. Program committee members must disclose any conflicts of interest through OpenReview (or, for reviewers, by notifying the cognizant Area Chair; for Area Chairs, by notifying the cognizant Senior Area Chair; etc). Any program committee member who is conflicted with a paper due to close ties to the authors of the paper cannot participate in deciding the fate of that paper. Program chairs are not exempt to this rule. In case a program chair is conflicted, the non-conflicted program chairs must make the decisions without the involvement of the conflicted program chair.

LLM use. Reviewers must follow the Policy for LLM use in reviewing.

Reviewers and Area Chairs that are also Authors

If reviewers or area chairs who are also authors neglect their reviewing or meta-reviewing duties, all of the submissions on which they are an author might be desk rejected, regardless of the merit of the submission. Examples of neglect of duties include failure to communicate in a timely manner, failure to meet the deadlines (without appropriate prior communication and arrangements with the relevant Area Chair or Senior Area Chair), violation of the Policy for LLM use in reviewing, and submission of low-quality reviews or meta-reviews.

Reporting and Enforcement

If you believe someone may be engaging in unethical conduct, please notify ICML by filling out the Ethics Violation Reporting form (link forthcoming; in the meantime, please report suspected violations to program-chairs@icml.cc).

All suspected unethical behaviors will be investigated by program chairs, integrity chair, or the ICML Oversight Committee, and individuals found violating the rules may face sanctions and/or have their submissions rejected. We will also collect names of individuals who are found to have violated ethics standards; if individuals representing conferences, journals, or other organizations request this list for decision-making purposes, we may make this information available to them.

Desk Rejection

Violation of requirements outlined in the Call for Papers, Author Instructions, and Peer-review Ethics may be grounds for desk rejection of a single submission or of all the submissions by the same author.

Grounds for desk rejection of an individual submission include (but are not limited to) the violations of:

  • Page limit.
  • Paper formatting.
  • Author anonymity.

Grounds for desk rejection of all the submissions by the same author include (but are not limited to):

  • Prompt injection (regardless whether all authors knew about it).
  • Violation of concurrent submission policy (that is, not citing and appropriately discussing concurrent ICML submissions with an overlapping set of authors in the body of the paper).
  • Neglect of reviewer or meta-reviewer duties (this is not limited to reciprocal reviewers).
  • Interference with integrity of peer-review process, including collusion, declaration of unsubstantiated conflicts, and attempts to uncover the identities of reviewers or meta-reviewers.