Preparing for FAccT 2027: Important updates for Authors and Reviewers

First, an important reminder: If you currently do not have an OpenReview account, please make sure to create one without delay. Authors without an institutional affiliation should sign up as soon as possible, because profile verification may take a couple of weeks to process. For these authors, we recommend creating an account by October 16, 2026, to allow enough time for verification. Please do not leave this until the submission deadline!

Dear FAccT community,

We’re excited to share info about several updates we have made to the submission and reviewing processes for the 2027 edition of FAccT. In this blog post, we’ll explain what is changing, what these changes mean for authors and reviewers, and how they support a rigorous, constructive, and fair review process.

What’s new for FAccT 2027

This year brings four important updates: a revised policy on authorship and AI writing tools; clearer boundaries for AI assistance in reviewing; the introduction of Senior Area Chairs; and a revised process that brings author rebuttals together with paper revisions. Please read the 2027 Author, Reviewer, and Area Chair Guides carefully—the details matter, even if you have submitted to or reviewed for FAccT before.

1. An updated policy on authorship and AI writing tools

For FAccT 2026, our policy prohibited using LLMs to generate text for publications, while allowing disclosed assistance with formatting, grammar, or writing fluency. For 2027, we are moving from that prohibition to an approach that is more aligned with ACM’s Policy on Authorship: generative AI may be used to create content, provided that its use is appropriately disclosed and human authors remain fully responsible for the work. AI systems still cannot be listed as authors.

What does this mean in practice?

• Substantive uses must be transparently disclosed. These include generating ideas, drafting arguments from scratch or from an outline, analysing data, producing findings, or generating text from notes and prompts that ask a tool to synthesise selected sources. If your paper’s content or structure has been shaped in any way by a generative AI tool, this likely qualifies as a substantive use.
• Disclosures should identify the tool and model version, explain the task it performed, and, where applicable, describe the extent of human review and/or validation of the approach. If the use of generative AI was core to the paper’s thesis, method, arguments, or conclusions, its role should be described in a Methods section.
• Responsibility cannot be delegated. Authors remain accountable for the correctness, originality, integrity, and presentation of all submitted content—including AI-assisted text, code, data, figures, analyses, and citations.

The Generative AI Usage Statement remains mandatory for every submission, and disclosure of any GenAI use, including substantive use, should be made here. Include it in the endmatter of your anonymised paper, whether or not you used generative AI. The statement does not count against the content page limit, and failure to disclose generative AI use is grounds for desk rejection.

There is an important distinction here: ACM generally exempts routine assistive AI use (e,g, basic spelling, grammar-checking, prose polishing, and translation) from its disclosure requirement, but FAccT still requires every submission to include reporting of substantive and assistive AI use in the usage statement. At submission, use that anonymised endmatter statement rather than an identifying Acknowledgements section.

Disclosure helps reviewers understand how a tool contributed to the work; it does not replace assessment of the paper’s scholarly contribution, methodological quality, evidence, or ethical implications. Reviewers should assess submissions on their merits, not treat polished prose or unfamiliar phrasing as evidence of inappropriate AI use. Specific, material integrity concerns should be raised confidentially with the Area Chair and Program Chairs.

2. Clearer boundaries for AI assistance in reviewing

The core ACM and FAccT principle has not changed: reviewers must do the reviewing themselves. The 2026 guide already prohibited LLM-written reviews, while permitting limited assistance with the language and formatting of a review the reviewer had written. The 2027 guide makes the boundary more explicit and strengthens the accompanying confidentiality guidance.

You may not use an LLM to undertake or substantively facilitate the assessment of a FAccT submission. This includes asking it to summarise, analyse, critique, assess, or evaluate any part of the paper. Checking an AI-generated assessment afterward does not make that use acceptable: scholarly evaluation is a professional responsibility that belongs to the reviewer.

Limited assistance with presentation remains permitted. After writing your own review, you may use a tool solely to improve its readability—for example, to check grammar, spelling, punctuation, fluency, formatting, or translation. Before uploading or pasting any review text into a third-party tool, you must remove all confidential content and anything that could identify the submission, authors, or reviewer. You must also check the returned text for invented claims, unintended changes, or other spurious additions.

Do not upload a submission, or any fragment of it, to a third-party AI interface or website outside the review platform. This ensures protection of confidentiality and the intellectual property of FAccT authors.

3. Introducing Senior Area Chairs

For 2027, we are adding Senior Area Chairs (SACs) as an operational bridge between Area Chairs (ACs) and Program Chairs (PCs). Their role is to support scientific quality, consistency, and fairness across a conference track.

FAccT brings together distinct methodological and disciplinary traditions. In that setting, careful judgments about individual papers can still produce uneven standards across different groups of submissions. The SAC role creates a structured opportunity to compare decisions across AC portfolios and identify inconsistencies, rather than leaving that work principally to the Program Chairs during calibration.

SACs will contribute in four ways:

• Mentorship and oversight: supporting ACs in meeting deadlines, actively moderating reviewer discussions, and producing thorough, constructive, well-grounded meta-reviews.
• Decision calibration: comparing borderline, sharply divergent, or contentious cases across AC pools to promote consistent acceptance thresholds and mitigate individual reviewer or AC bias.
• Conflict and escalation management: responding to AC withdrawals, ethical concerns, or serious disagreements, including by providing additional evaluations or seeking further informed opinions.
• Final programme curation: making high-level recommendations to the Program Chairs and participating in final calibration.

This is intended as a quality, consistency, and resilience mechanism—not simply another administrative layer. ACs already carry substantial responsibilities for reviewer assignments, review quality, discussion, and recommendations. SACs distribute supervisory work and provide an additional route for addressing difficult cases or breakdowns in the process. The aim is more comparable decisions, earlier quality control, and clearer, more substantive explanations for authors.

4. Bringing rebuttals and revisions together

FAccT 2026 introduced a revise-and-review process, but it also retained a separate, short rebuttal period before the first-round decision. That rebuttal focused on correcting misunderstandings or factual errors, rather than introducing new data or debating substantive disagreements. Papers receiving a Revise decision then proceeded to a later revision and re-review stage.

For 2027, there will no longer be a separate rebuttal stage before the initial decision. Instead, papers receiving Revise (minor) or Revise (major) are revised after the initial decision, and authors submit their revised paper together with a rebuttal explaining how they addressed the prioritised feedback. Revised papers may be up to 15 content pages, excluding references and endmatter.

Here is how the updated process will work:

1. Initial review and discussion. Reviewers assess the paper and discuss their evaluations under the AC’s moderation.
2. Initial decision. Authors receive an Accept, Revise (minor), Revise (major), or Reject decision, together with reviews and a meta-review explaining the reasoning. For Revise decisions, the meta-review identifies and prioritises the issues to address.
3. Combined revision and rebuttal. Authors receiving either Revise decision submit an updated paper and a response explaining how they addressed the feedback. Revised papers may use an additional content page, bringing the limit to 15 pages, excluding references and endmatter.
4. Re-review and final decision. The original reviewers and AC assess the revised paper and rebuttal, discuss the changes, and update their assessments. Final calibration involves PCs and SACs, leading to an Accept (with minor revisions) or Reject decision. Papers accepted at the initial stage proceed to camera-ready preparation, with minor changes if requested.

The aim is to connect the author response directly to the revised scholarship: reviewers can consider both the explanation of how concerns were addressed and the changes made to the paper itself.

A few clarifications on frequently asked questions

Our Program Chair inbox has already been receiving a steady flow of emails, so we thought it might be helpful to provide general reference answers to some of recurring queries.

• Do all authors and co-authors on my FAccT 2027 submission need to sign up to be reviewers? Yes. We require all authors of every paper to sign up to review for the conference, though not all authors will be required to review. Of those who fill out the form, only qualified individuals will be added to the roster of reviewers.
• I am really interested in volunteering to review for FAccT 2027. Can you sign me up? If you’re interested in reviewing for FAccT, please use the reviewer form linked here https://facctconference.org/2027/cfp.html to sign up.
• What is the expected reviewer load for FAccT 2027 and can I request a reduced load given my other commitments? At the moment, the expected reviewer load for FAccT 2027 is 4 to 5 papers, and a reduced load of 3 papers is possible. Keep in mind that these are estimates, at this point, and that the expected reviewer load may be a bit more or less, depending on the number of reviewers in the final roster.
• I have a submitted a short article to a journal that contains content similar to my FAccT 2027 submission, but, while my FAccT submission repeats some of the content from the journal article, it is substantially different, building on the journal article but adding mostly new content. Does this count as a dual or concurrent submission? A FAccT submission that is substantially different from a journal article currently under review would not ordinarily be treated as an impermissible concurrent or dual submission merely because the two papers share some content. The relevant test is whether the FAccT paper is identical or substantially similar to the manuscript under journal review. Moreover, FAccT 2027 expressly permits a conference-length version of a paper already under review at a journal where the author selects the non-archival submission option and ensures that the parallel submission complies with the journal’s own policies.

Working together to make our review process the best possible Thank you to everyone preparing submissions and contributing their time and expertise to reviewing and chairing for FAccT 2027. Thoughtful and constructive reviews, active discussions, clear meta-reviews, and careful author responses and revisions all help FAccT continue to strengthen the quality of its outputs and its interdisciplinary mission.

We look forward to reading your work and building the FAccT 2027 program together!