Skip to yearly menu bar Skip to main content


ICML 2026 Area Chair Instructions

 

Thank you for serving on the ICML 2026 program committee as an Area Chair!

Immediate TODO items:

  • Please make sure you are available in the key reviewing periods, and are able to meet the AC deadlines (see key dates below).
  • Please read the entirety of the area chair instructions (this document), the reviewer instructions, as well as these fun and informative Peer Reviewing memes.
  • Please carefully review the ethical conduct and use of generative AI policies (below).
  • Please make sure your OpenReview profile is up-to-date, including conflicts of interest.

Responsibilities of Area Chairs

The responsibilities of an Area Chair (AC) for ICML are as follows:

  • Indicate your areas of research expertise, and “bid” on submissions to meta-review.
  • Check your meta-reviewing assignments and notify your Senior Area Chair (SAC) of any problems (e.g., conflicts of interest).
  • Check that each assigned submission is assigned four qualified reviewers (we may revise this), and adjust assignments as necessary.
  • Ensure that each assigned submission receives at least three high-quality reviews, each written in a respectful and professional manner.
  • Recruit emergency reviewers as needed.
  • Engage reviewers in discussions of submissions.
  • Carefully read reviews and discussions to write informed meta-reviews and make accept/reject recommendations.
  • Be responsive throughout the reviewing period, especially to your assigned reviewers and SAC.
  • Provide ratings of the reviewers for your assigned submissions.

Key Dates

  • Bidding period: January 27–Feb 2, 2026
  • Full paper submission deadline: January 28, 2026
  • Submission assignment period: February 3–11, 2026
  • Check/adjust reviewer assignments: February 6-10, 2026
  • Reviewing period: February 12–March 12, 2026
  • Deadline for reviews: March 12, 2026
  • Emergency reviewing period: March 13–23, 2026
  • Authors’ response and discussion period: March 24–April 7, 2026
  • Deadline to acknowledge authors’ response: April 3, 2026
  • AC-reviewer discussion period: March 31–April 12, 2026
  • Meta-review deadline: April 13, 2026
  • AC-SAC discussion period: April 14–April 18, 2026
  • Author notification: April 30, 2026

Ethical Conduct for Peer Review

Members of the program committee, including reviewers and ACs, are expected to follow standard ethical conduct for peer review. In particular, ICML prohibits:

  • use of privileged information (e.g., information and discussions about submissions) for any purpose other than reviewing
  • use of Generative AI tools in reviewing that violates the reviewers’ LLM policy (as described below). Use of Generative AI tools in the AC workflow is not permitted in any form;; 
  • all forms of collusion, whether explicit or tacit (e.g., an arrangement between authors and reviewers, ACs, or SACs to obtain favorable reviews).

Use of Generative AI tools (such as LLMs) for reviewing is permitted within the framework of the LLM review policy. In particular, reviewers following Policy A are not allowed to use any Generative AI tools in the review process, while there are acceptable forms of usage under Policy B. The reviewers' assigned LLM policies will not be available to ACs, so treat all the reviews and reviewers uniformly. Of course, report any issues that might suggest full delegation of review writing to Generative AI, such as hallucinations, to program chairs and integrity chair via our Ethics Violation Reporting form (this is not allowed under either policy). Use of Generative AI tools in the AC workflow is not permitted in any form.

Please also see the ethics guidelines of ICML 2026.

If you believe someone may be engaging in unethical conduct, please notify ICML via the Ethics Violation Reporting form.

All suspected unethical conduct will be investigated by ICML’s oversight committee. Individuals found violating the rules may face sanctions, have their own submissions rejected, etc.

 

Additional Details of the Reviewing Process for ACs

See the Reviewer Instructions for details of the reviewing process. Below are additional items specific to ACs.

Your Senior Area Chair

Your work as an AC is overseen by a Senior Area Chair (SAC). This SAC is your first point of contact for any issues that arise throughout the reviewing process. The name and email of your SAC will be visible at the top of the AC console in OpenReview.

Bidding

During the bidding period, you will indicate “bids” on (abstracts of) submissions that you are interested in (and qualified for) handling. This process is done within OpenReview. These bids will be used, in part, to help with the assignment of submissions to ACs.

Checking/Adjusting Reviewer Assignments

After submissions are assigned to ACs and reviewers, you will be able to check and possibly adjust the reviewer assignments for submissions they are handling. The reviewer assignments will attempt to balance a number of concerns, so please do not make more than two reviewer substitutions per submission. If additional substitutions are absolutely necessary, please obtain approval from your SAC.

Ensuring Review Quality and Rating the Reviews

Each submission must receive at least three high-quality reviews, each written in a respectful and professional manner. Please check the reviews as they come in for your assigned submissions. If necessary, you can ask the reviewer to revise their reviews (e.g., to add additional justification, clarify points or questions, adjust the tone/language).

(Emergency) Reviewing Period

It is likely that some reviewers will not have their reviews submitted by the deadline. Get in contact with the reviewers---ideally before the reviewing deadline---to ascertain if they will eventually submit their reviews before the author response period begins, or if you need to recruit emergency reviewers (perhaps yourself, if absolutely necessary). Use your best judgement or ask your SAC for advice.

Instructions for recruiting emergency reviewers will be given closer to the start of the emergency reviewing period.

Try to not end up with too many (say, 5+) reviewers for any single submission, as this can be a burden to authors. However, if you do end up in this situation, please make sure authors and reviewers know that authors cannot be expected to respond to all reviews. If possible, let the authors know which reviews should be prioritized, and keep this in mind when you form your meta-review.

Discussions

After the authors' responses are posted, make sure that your reviewers read them (and acknowledge that they have done so) by the stated deadline. New this year, the reviewers are also required to provide a final justification which describes the reasoning behind their final recommendation, accounting for both the paper and the rebuttal. Please work with the reviewers in ensuring that this is entered in a timely manner, and appropriately reflects the author-reviewer discussions.

Make sure the protocol of the Author-Reviewer discussions is followed by the authors and reviewers. Like last year, we are limiting the number of rounds of communication between authors and reviewers.

Additionally, encourage your reviewers to discuss submissions with each other (and with you) by posting and responding to notes on OpenReview (visible only to Reviewers/ACs). This is especially important if there are important contradictions between reviews, or if any key aspects of the paper remain unclear to a reviewer. AC-Reviewer discussions will not be visible to authors, so encourage reviewers to appropriately update their reviews after discussions for the benefit of the authors.

Meta-Reviews

Carefully read the reviews and discussions to make accept/reject recommendations and compose  informed meta-reviews. Each meta-review should summarize the key points that justify your accept/reject recommendation, whether from the reviews themselves or the (anonymized) discussions.

The review form this year is new and has specific items that reviewers are required to complete. Your recommendations and meta-reviews should be informed by the review content, as opposed to just relying on average (or other aggregate) numerical scores.

Please submit the recommendations and meta-reviews for all of the submissions you are handling by the deadline.

For the next week or so after the meta-review deadline, your SAC may reach out to you about your meta-reviews (e.g., for clarifications, possible adjustments/calibration).

Rating the Reviewers

Please provide ratings of the reviewers for your assigned submissions. This feedback is useful for highlighting reviewers who go above-and-beyond in their reviewing work, and also to identify reviewers who are consistently delinquent or shirk their responsibilities even after repeated reminders. (We expect the vast majority of reviewers to "meet expectations".) If you have any questions or concerns about providing these ratings, please ask your SAC for advice. 

 

AC Instructions to Edit Reviewer Assignments

These instructions are based on those used for ICML 2025.

Assigning emergency reviewers for a paper

You should first contact a potential reviewer via email to check if they are willing to be an emergency reviewer (with a meaningful deadline so that you can ensure 3 high quality reviews sufficiently ahead of the beginning of the authors’ response period of March 24th). In this email, you can also communicate additional information about the submission. Do this before sending the invitation via OpenReview! (Unfortunately, due to a limitation in OpenReview, the "task" will still display a deadline of March 12, so it is important to communicate the actual deadline with the potential emergency reviewer.)

You can add emergency reviewers by picking from the pool of available emergency reviewers, or adding a new external reviewer to the pool (which requires the SAC intervention for the latter).

You can see the pool of available emergency reviewers in the "Modify Reviewer Assignments" interface (see "Modifying the reviewer assignments for a paper" in the AC instructions to access it) by clicking on the checkbox “Show reviewers with fewer than max assigned papers”.

Emergency reviewers are identified via the “Custom Max Papers” field that we have modified accordingly. If the “Assignments” field is strictly below the “Custom Max Papers” field, then this reviewer is available for emergency reviews, and you can add them by clicking the “Invite Assignment” button.

Note that OpenReview does not prevent you from inviting a reviewer even if they have reached their quota. Please do not invite someone with a full quota as we have carefully designed the emergency reviewer pool via this “custom max papers” field.

Clicking the “Invite Assignment” button will send them an automatic email inviting them to review the paper and they need to accept the invitation via a link in the email before they become assigned to the paper. As you cannot personalize this invitation email, please also send them a direct email to communicate with them about the emergency reviewing situation. Note that this email will be signed with your name. (It is fine for the reviewer to know your identity and vice-versa.)

Finally, note that if you struggle to find emergency reviewers for a paper, you can also add yourself as a reviewer as a last resort by asking your SAC to invite you as an external reviewer for it.

Modifying the reviewer assignments for a paper

Go to your Area Chair Console: https://openreview.net/group?id=ICML.cc/2026/Conference/Area_Chairs

Click on the "Modify Reviewer Assignments" link at the top, and then click on a paper on the left side to bring up the assigned reviewers.

To un-assign a reviewer, click the "Trash" icon.

To filter out reviewers that have already reached their full quota, check "Only show reviewers with fewer than max assigned papers".

You can sort reviewers according to various features, including Affinity and Bids. We recommend changing the default sort from "Assignment" to something more informative, such as "Affinity Score", to ensure that the most relevant reviewers appear at the top.

To add a reviewer, select from among the unassigned reviewers and click "Assign" / "Invite Assignment".

 

Ethics Review Instructions

As part of the ethics review process at ICML, reviewers were asked to flag papers for which they have identified an ethical concern. Categories of ethical issues, including an illustrative example of expectations for each category,  can be found at https://icml.cc/Conferences/2026/ResearchEthics . 

Your job is to triage these flags. You will be asked to select one of the following categories for each paper. 

  • No action necessary. 
    • This paper raises no ethical concerns (i.e. no reviewers flagged it and you agree with that decision)
    • The ethical issue requires no further discussion. (i.e. a reviewer flagged the paper but, in your judgment, the issue raised is not serious enough to require further discussion or remediation). This may happen if there is not a direct connection between the paper’s contribution and the issue. For example, it is possible that a reviewer flags a purely theoretical paper on the grounds that the theoretical advancement could potentially be used to further ethically problematic use cases. If the scientific contribution could equally apply to many other applications, this can be dismissed without further discussion. Other flags based on a tenuous connection to ethical concerns can similarly fall into this category.
  • AC involvement. 
    • The remediation process is being handled by the reviewers and authors. We expect that for many cases where an issue is raised, the authors will be able to appropriately address the issue by providing a more robust discussion of the risks or steps they have taken to ensure compliance with, e.g., their local legal standards. If you feel confident that you and the reviewers can provide reasonable feedback about what constitutes an acceptable discussion of issues and can exercise reasonable judgment of whether the  issue has been adequately addressed, select this option.
    • The ethical issue is not being adequately addressed by the reviewers and authors but you feel capable of managing the issue, for example by engaging the authors on the need for more discussion on the issue.
  • Ethics review needed. 
    • You recognise that the ethical issue is not being addressed appropriately by the reviewers and authors, and you do not feel capable of managing the issue yourself.
    • You feel that you need additional input on how to appropriately mitigate the issue or whether there is an issue at all. 
    • In this case, the paper will proceed through the full ethics review process. You will need to select the most relevant area of concern so that the paper can be distributed to ethics reviewers with relevant expertise. 
  • ​​​​​Integrity review needed.
    • You believe that the integrity issue (e.g. plagiarism, AI-generated paper, etc.) requires review by the Program Chairs

Ethics review process

The paper will be forwarded to ethics reviewers with relevant expertise. You can expect to receive two ethics reviews. Given the tight timeline for these reviews, we will forward them to you as they are completed and will try to get them back to you within the rebuttal window. 

Ethics reviewers will provide feedback on the specific issues identified and have been instructed to, where possible, give constructive feedback on how the issue can be remediated. If the ethics reviewers determine that there are reasonable mitigations available, it will be your job to work with the authors to be sure that these are enacted. 

In some cases, it is possible that the ethics reviewers identify an issue that is beyond remediation. In this case, one of the Ethics Chairs will serve as a third reviewer and provide a recommendation on how to proceed. Ultimately,  the decision to accept or reject the paper will be made by the Program Chairs with input from you and the reviewing ethics chair.