AI Acknowledgement

Journals and conferences / Association for Computational Linguistics (ACL) and ACL Rolling Review (ARR)

ACL policy on AI writing assistance (ARR)

ACL's policy lets you use generative AI only for listed uses: polishing language, predictive typing, literature search, low-novelty text, and new ideas you then develop yourself. Any other use counts as inappropriate. Use for writing or coding must be disclosed in question E1 of the Responsible NLP Checklist, with details in the Acknowledgements section; grammar checkers, spell checkers and predictive keyboards don't need disclosure. This applies to papers submitted through ACL Rolling Review, which feeds venues such as EMNLP, NAACL and EACL.

Independent guide, not an official Association for Computational Linguistics (ACL) and ACL Rolling Review (ARR) page: we aren't affiliated with or endorsed by Association for Computational Linguistics (ACL) and ACL Rolling Review (ARR). Summarised from its official guidance, last checked 21 November 2026. Rules can change, so your call for papers wins if it says something different · See the full rules

Write yours

Association for Computational Linguistics (ACL) and ACL Rolling Review (ARR) doesn't publish a fixed template, so this uses a general format that covers what it asks for.

Did you use AI?

Your university or journal

Which AI tools?

What did you use it for?

What did you do with what it gave you?

More details · prompts, date, versions
Authors
Advertisement

Association for Computational Linguistics (ACL) and ACL Rolling Review (ARR)'s own wording

Section 3 was written with inputs from ChatGPT.

Quoted from Association for Computational Linguistics (ACL) and ACL Rolling Review (ARR)'s guidance.

Association for Computational Linguistics (ACL) and ACL Rolling Review (ARR)'s own examples

Quoted word for word. Pick the one closest to what you did.

  • ACL's own example of an Acknowledgements disclosure

    Section 3 was written with inputs from ChatGPT.

Source: Association for Computational Linguistics (ACL) and ACL Rolling Review (ARR)'s guidance.

Worked example

Illustrative example: Acknowledgements. We used ChatGPT to suggest wording for the background part of Section 2, which describes well-known concepts; we checked it for accuracy and added citations. We used GitHub Copilot while writing our data processing scripts; this is also noted in the code README. All content was checked by the authors.

The rules at a glance

When you need one
Disclose any use of generative AI to create content, whether for the research or for review rebuttals. ARR says use for writing or coding, and its scope, must be disclosed in the Responsible NLP Checklist (question E1). You don't need to disclose tools that only help with proofreading (grammar or spell checkers) or short-form input help (predictive keyboards, typing suggestions). For low-novelty text, say where generated text was used. For new ideas from a model, disclose that models were used this way.
Framework
ACL Policy on Publication Ethics, 'Guidelines for Generative Assistance in Authorship' (cases a-f), applied by ARR plus Responsible NLP Checklist question E1.
Where it goes
Answer checklist question E1 (in the submission's Responsible NLP Checklist form) and give details in the paper's Acknowledgements section. For coding assistance, details may also go in README files.
Referencing
Don't list the AI tool as an author. If generated text copies existing work, cite both the source of the text and the source of the idea(s).
Must include
  • checklist E: whether you used AI assistants (e.g., ChatGPT, Copilot) in research, coding, or writing
  • checklist E1: if yes, a section number or your elaboration directly in the checklist response; if no, a justification
  • the use and its scope for writing or coding
  • for low-novelty generated text: where it was used, that it was checked for accuracy, and relevant citations
  • for model-suggested ideas: a statement that models were used this way, plus citations to prior work found by a literature search
Also note
  • generative AI tools may not be listed as authors; ACL does not consider a model able to meet co-authorship requirements
  • any use not on ACL's list of appropriate uses is considered inappropriate
  • authors are fully responsible for the correctness of methods, results and writing, and should check for plagiarism of both text and code
  • authors are encouraged to use ethically-sourced and open models when they need LLMs, including for rebuttals
  • submissions trying to manipulate machine readers, e.g. with prompt injections, may be desk rejected; camera-ready papers with such attempts may have to be retracted
  • submissions that systematically fail to answer relevant checklist questions with sections or justifications will be desk-rejected
  • reviewers must write the content and argument of their review themselves; using generative AI for the first draft is not permitted (this also applies to meta-reviews)
  • reviewers may use writing help to paraphrase a review, and tools that help check proofs
  • reviewers and editors must not upload a submission, or their review, to a non-privacy-preserving generative tool
  • reviewers who rely on generative models in breach of policy can be sanctioned as 'highly irresponsible', including a temporary ban from (re-)submitting at ACL venues

Questions about Association for Computational Linguistics (ACL) and ACL Rolling Review (ARR)'s rules

Can I use ChatGPT to write my ACL / EMNLP paper?
Only for the uses ACL lists. Polishing your own text, literature search, and low-novelty text about known ideas are fine (with checking and citations). Letting a model produce both new ideas and the text for them is not, because ACL doesn't accept a model as a co-author. Any use not on the list counts as inappropriate. Source
Do I need to disclose Grammarly or a spell checker?
No. Tools that only help with proofreading, like grammar or spell checkers, and predictive keyboards don't need to be disclosed. Source
Where do I put the AI disclosure for an ARR submission?
Answer question E1 of the Responsible NLP Checklist, and put the details in the Acknowledgements section. For coding help, details may also go in README files. Source
What is question E1 on the Responsible NLP Checklist?
Question E asks whether you used AI assistants (e.g., ChatGPT, Copilot) in your research, coding, or writing. E1 asks whether you included information about that use. For yes, give a section number or explain in the checklist; for no, justify why not. Source
Can ARR reviewers use LLMs?
Not to write the review. Reviewers must write the content and argument themselves, and can't use AI for the first draft. They may use writing help to paraphrase a review, and must not upload a paper or their review to a non-privacy-preserving tool. Source
When is the next ARR deadline?
When we checked on November 21, 2026, the ARR dates page listed cycles only through October 2026 (submission October 12, 2026), with NAACL 2027 commitment on December 23, 2026. Later cycles weren't listed yet, so check the page for updates. Source

If you didn't use AI

Checklist question E1: if you answer no, you must give a justification. ACL doesn't give example wording.

Sources

Your call for papers wins if it says something different. Spotted a change? Tell us.