Journals and conferences / IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
CVPR 2027 LLM policy
CVPR 2027 lets authors use LLMs and other AI tools, but every submission must include a section titled "LLM Usage Statement" after the main text and before the references. It must disclose any use, including editing and grammar, name the tools, say which parts were affected and how much; if you used none, you state that. Papers without the statement will be desk-rejected, and papers citing non-existent work will be rejected. Reviewers may only use the official LLM assistant inside OpenReview.
Independent guide, not an official IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) page: we aren't affiliated with or endorsed by IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). 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
IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) doesn't publish a fixed template, so this uses a general format that covers what it asks for.
IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)'s own wording
LLM Usage Statement. We used [tool name(s)] for [nature of use] in [parts of the paper or research affected, e.g. text, figures, code, experiments, related work or ideation]. [Brief description of the extent of use.]
Summarised from IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)'s guidance.
Worked example
Illustrative example: LLM Usage Statement. We used GPT-5 to fix grammar and shorten sentences in Sections 1 and 4, and GitHub Copilot to help write data loading and plotting code. No AI tool was used for ideation, experiments or related work. We checked all edits and code ourselves.
The rules at a glance
- When you need one
- Always. Every submission needs an LLM Usage Statement, even if no AI was used. It must cover any use of LLMs or other generative AI tools in the research or the writing, including for editing and grammar.
- Framework
- CVPR 2027 Author Guidelines, Section 2 'LLM Policy' and Section 6 LLM FAQs, aligned with IEEE's AI-generated content rules.
- Where it goes
- A section titled "LLM Usage Statement" after the main text and before the references. It doesn't count toward the page limit and must not reveal author identity. In the camera-ready version, AI-generated content must be disclosed in the Acknowledgments (the statement may be moved there, de-anonymized).
- Must include
- any use of LLMs or other generative AI tools in the research or the writing, including for editing and grammar
- name of the tool(s) used (a generic technical description during review if naming a non-public tool would break anonymity, full name in camera-ready)
- the parts of the paper or research affected (for example text, figures, code, experiments, related work or ideation)
- a brief description of the nature and extent of the use
- if none used: the sentence "No LLMs or generative AI tools were used in this work"
- Also note
- papers without an LLM Usage Statement will be desk-rejected; a statement that leaves out significant information may lead to desk rejection or referral to the ethics committee
- papers that cite non-existent work, or contain obvious factual fabrications, will be rejected and may be rejected without review
- an LLM should not be listed as an author (IEEE authorship criteria)
- "An LLM did it" is not a defense against plagiarism, fabrication or inaccuracy
- prompt injection (hidden instructions to manipulate reviewers or automated tools) is a collusion attempt; such papers will be desk-rejected and authors may face more sanctions
- papers may be processed by conference-approved LLM tools during review; there is no opt-out
- reviewers may only use the official LLM assistant inside OpenReview; any other LLM or chatbot in any part of reviewing is prohibited
- reviewers can't share the paper, supplementary material, review, rebuttal or discussion with any other LLM (not even to translate a review); grammar checkers based on LLMs are allowed
- reviews are checked for LLM policy violations; violating reviewers' own papers may be desk rejected at the PCs' discretion
Questions about IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)'s rules
- Can I use ChatGPT for my CVPR 2027 paper?
- Yes. Authors may use LLMs and other AI tools, but you're fully responsible for everything you submit and must be able to explain, reproduce and defend it. You must describe the use in your LLM Usage Statement. Source
- Do I need to disclose grammar fixes at CVPR?
- Yes. The LLM Usage Statement must disclose any use of LLMs or other generative AI tools, including for editing and grammar. Source
- What if I didn't use any AI?
- You still need the statement. Write "No LLMs or generative AI tools were used in this work". Papers without an LLM Usage Statement will be desk-rejected. Source
- What happens if my paper has a hallucinated reference?
- Papers that cite non-existent work, or contain obvious factual fabrications, will be rejected and may be rejected without review. Source
- Can CVPR reviewers use ChatGPT?
- No. The only LLM allowed in reviewing is the official assistant inside OpenReview, and using it is optional. Sharing the paper or review with any other LLM is not allowed. Grammar checkers based on LLMs are allowed. Source
- When is the CVPR 2027 deadline?
- Paper registration was November 10, 2026 and the paper submission deadline was November 16, 2026 (Anywhere on Earth). Supplementary materials are due November 23, 2026. CVPR says these dates won't be extended. Source
If you didn't use AI
No LLMs or generative AI tools were used in this work
More help
Other journals and conferences
- ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD)
- Association for Computational Linguistics (ACL) and ACL Rolling Review (ARR)
- Conference on Neural Information Processing Systems (NeurIPS)
- Elsevier
- Frontiers
- IEEE
- International Conference on Learning Representations (ICLR)
- International Conference on Machine Learning (ICML)
- International Joint Conference on Artificial Intelligence (IJCAI)
- MDPI
- PLOS (Public Library of Science)
- SAGE Publishing
Sources
- https://cvpr.thecvf.com/Conferences/2027/AuthorGuidelines
- https://cvpr.thecvf.com/Conferences/2027/ReviewerGuidelines
- https://cvpr.thecvf.com/Conferences/2027/CallForPapers
- https://cvpr.thecvf.com/Conferences/2027/Dates
Your call for papers wins if it says something different. Spotted a change? Tell us.