ICML 2026 New Guide to Avoid Pitfalls: Participation is Not Mandatory, Original Scripts Will Be Published, and Peer Review Has a Limit
Posted Time: 2025 November 6 15:03
AuthorThe New Intelligence of Science and Technology
Image 0
Machine Mind Report
Machine Mind Editor Department
Attention to those who want to submit to the ICML 2026 top conference, there are new rules for paper submission, including but not limited to:
Authors whose papers have been accepted have the option of whether or not to participate in the conference. The originally submitted version of the accepted paper will be made public. There is a limit to the number of papers that will undergo mutual
Image 5
ICML 2026 Call for Papers Guidelines
ICML 2026 will be held in Seoul, Korea between July 7th and 12th, 2026. All papers will be reviewed with a double-blind review system, and the accepted papers will be presented at the conference.
Contributions must be submitted in the form of a single document, and the specific requirements are as follows:
The main text should be no more than 8 pages long; the references, impact statement, and appendix sections have no page limit; there is no separate deadline for supplementary materials.
Note: After the paper is accepted, the author can add an additional page to the main text in the final version.
Policy Change: Authors Can Choose Whether to Attend the Conference After Their Papers Are Accepted
After receiving the paper acceptance notice, the author needs to decide by the specified date whether to attend the conference in person to present the paper or only submit the paper to the conference proceedings (without a live presentation).
Regardless of the author's choice of submission method, all accepted papers will be treated equally in the conference proceedings and will be eligible for the ICML awards.
For papers that are only selected for inclusion (without presentation), at least one author must complete virtual registration.
Policy Change: In addition to the final version, the original submission of the published paper will be available
For all accepted papers, in addition to the final draft version, the following contents will be published by the conference:
Original submission version (including supplementary materials); anonymous review comments; meta-reviews; rebuttal; and records of discussion between reviewers and authors.
In addition, authors of rejected papers may choose whether to make their original submission, review comments, meta-reviews, rebuttals, and review discussion records public.
Moreover, ICML 2026 also has other requirements:
The same person cannot be designated as a reviewer for too many papers at the same time; policies for using AI tools in the review process will be updated (specific details will be announced later).
Any papers that do not meet the above requirements will be rejected directly without entering the review process.
Theme for submission
ICML 2026 Call for Papers topics include:
General machine learning (active learning, clustering, online learning, ranking, supervised learning, semi-supervised and self-supervised learning, time series analysis, etc.) Deep learning (architecture, generative models, theory, etc.) Evaluation (
Important Dates
The first is the submission deadline for each part of the paper:
Submission website opening date: January 8, 2026 (This is also the suggested deadline for creating an OpenReview account. Authors who do not yet have an OpenReview account should register in advance to ensure timely account activation.) Abstract subm
Submit website: https://openreview.net/group?id=ICML.cc/2026/Conference
Please note that the deadline for abstracts and paper submission is strictly enforced and there will be no extension of any kind.
Abstract / Title / Author Identity Modification:
Authors must complete the title, author list, and abstract before the abstract submission deadline; the title and abstract can be modified before the full paper submission deadline. However, if the placeholder abstract is significantly rewritten duri
Review Mechanism
Mutual review requirement: ICML requires authors to participate in the review process with two requirements
Peer review requirements for each paper:
All submissions must have at least one author agree to serve as a reviewer for ICML, and the designated author should possess the qualification defined in the FAQs for Peer Review. New rule this year: one author can only be designated as a mutual rev
Mutual review requirements for each author
If an author has submitted 4 or more papers, they must serve as an ICML reviewer (can serve as a mutual review reviewer for two of their papers); if there is a shortage of reviewers, the project chair can lower this threshold to 3 papers with advance
Note: If the requirements for mutual evaluation are not met, the submission may be rejected directly. In addition, reviewers who fail to complete their reviews on time or in a perfunctory manner may also have their submissions rejected.
Double-blind Review Policy
All submissions must be anonymized and must not contain any intentional or unintentional information that reveals the author's identity. Authors are allowed to publish their papers on preprint platforms such as arXiv and present them during the revie
Prohibition of double/parallel submission
It is prohibited to submit the same paper or substantially similar content to multiple conferences or journals. Violations of this rule will result in the rejection of the paper or removal from the conference proceedings. However, papers presented at
Review Criteria:
The paper should report original, rigorous, and significant research findings in the field of machine learning; all arguments must be supported by reproducible experiments or solid theoretical analysis, and clearly explain the relationship and differ
Regulations for the Use of Generative AI:
Authors are allowed to use generative AI such as large language models to aid in writing or research, but they must take full responsibility for the content of their papers. LLM cannot be credited as an author. Any form of 'prompt injection' is prohi
Peer review ethics: Authors must adhere to standard research and review ethics, including:
No plagiarism. No hint injection. No promotion during the review period as submitting to ICML. No collusion with reviewers, ACs or SACs to seek favorable or unfavorable reviews.
It should be noted that those who violate the rules may face sanctions or rejected submissions, and their information may also be reported to other academic institutions.
Impact Statement:
Each paper must be accompanied by a statement of potential social impact, including ethical and future social impact. This statement should be placed in a separate section at the end of the paper (in the same place as the thanks, before the reference
If research only drives progress in the field of machine learning, you can use the following short template:
This paper aims to promote the development of machine learning. We believe that this research does not need to emphasize social impact in particular.
It should be noted that reviewers and ACs can mark papers as subjects of ethical review, and those with serious violations may be rejected due to ethical issues.
Summary: Authors submitting papers for acceptance are required to provide a plain language summary to facilitate communication of research significance to the public. The official will provide specific examples and templates.
Optional self-rating: Similar to previous years, authors with multiple submissions can voluntarily rank their papers in quality.
https://x.com/icmlconf/status/1986089104367308805
https://icml.cc/Conferences/2026/CallForPapers