All JournalsPhysics Magazine

Appropriate Use of AI Tools

Artificial intelligence tools are rapidly evolving, and APS Journals continue to observe and learn about their uses from our authors and reviewers.

AI tools may be used to assist with aspects of research, manuscript preparation, and peer review. We are committed to establishing clear boundaries where human judgment is required.

This policy governs the use of AI by authors during manuscript preparation and by reviewers during peer review.

  • Any AI used to conduct the actual research (e.g., data analysis) must, like any tool, be disclosed within the paper’s methods section.

  • AI assistance in preparing a submitted manuscript or conducting a peer review must follow the disclosure standards outlined below.

We will continue to monitor AI’s capabilities and limitations, updating these guidelines as the technology evolves.

Authors

AI tools cannot replace authors’ expertise, judgment, or accountability. An AI program cannot be held accountable and cannot be listed as an author. Authors are fully responsible for accuracy and must review and refine all AI-generated content.

Disclosure

Authors must keep a record of any AI use and disclose any substantive uses in the submitted paper (see FAQ for more details).

Authors should state:

  • AI tool name and version
  • How the AI assisted
  • How the authors directed and verified the AI output

The editors may request more documentation to support transparency if concerns emerge. Authors are not required to disclose the use of AI to polish, condense, or otherwise lightly edit the text.

Rights Protection

Authors must check the AI tool terms and conditions to ensure the tool claims no rights over their content and imposes no usage limits. As terms and conditions change over time, please revisit these terms regularly, especially regarding privacy and data protection.

AI can be used for data visualizations, but authors are responsible for verifying the accuracy of all AI-generated figures or graphics. Authors must ensure these images do not violate third-party copyrights, licensing agreements, or attribution requirements. Authors must continue to follow all APS journals Editorial Policies and Practices to ensure that the manuscript is original and that proper permissions have been obtained. Any images or non-textual media should be factual, scientifically correct, and accurately represent the content of the paper.

At this time, APS does not use AI-generated art for key images, cover images, or other images on its websites or in promotional materials unless they are from trusted third-party sources or are an allowed AI-generated data-visualization from the paper.

Reviewers

Reviewers may use AI to:

  • Organize reviewer thoughts and prioritize feedback
  • Improve clarity and tone or translate the review
  • Query AI as a learning tool

AI may support writing but cannot replace reviewer expertise and judgment. Reviewers are fully responsible for accuracy and must review and refine all AI-generated content.

Disclosure

If AI is used to help draft or edit a review beyond polishing, condensing, or otherwise lightly editing the text, disclosure is required. Reviewers should state, in a note to the editor:

  • AI tool name and version
  • How the AI assisted
  • How the reviewer directed and verified the AI output

Editors may request more documentation to support transparency.

Rights Protection

Reviewers must check the AI tool terms and conditions before use. Because review materials are confidential, AI tools may only be used if their terms of use state that uploaded manuscripts or reports will not be used as training data. Authors have not consented to their unpublished work being shared with AI systems, and unreviewed manuscripts may contain errors that could propagate through training datasets. As a result, uploading manuscripts and reports (in total or in part) to unrestricted AI tools is not permitted.

Reviewers must also ensure the tool claims no rights over their content and imposes no usage limits. As terms and conditions change over time, please revisit these terms regularly, especially regarding privacy and data protection. At all times, reviewers are responsible for the final report and must take care that mistakes made by the AI do not slip through. Failure to do so undermines the peer review process and may cause a publication ethics investigation.

Editors

AI tools present opportunities to support the peer review process but cannot replace editorial judgment and decisions. Editors may only use APS-managed and licensed AI tools, which are vetted to ensure that uploaded content is never used as training data. Within those tools, editors may use AI to assist with certain tasks, such as identifying potential reviewers, polishing or condensing text, or drafting online summaries.

Disclosure

If AI is used in editorial writing beyond polishing, condensing, or otherwise lightly editing the text, disclosure is required, including but not limited to:

  • AI tool name and version
  • How the AI assisted
  • How the editor directed and verified the AI output

FAQ

What constitutes substantive use of AI?

Substantive AI use includes, but is not limited to, assistance with:

  • Scientific reasoning or interpretation
  • Drafting or revising scientific claims, conclusions, or explanations
  • Literature synthesis beyond routine organization
  • Figure generation or image analysis
  • Derivations or calculations
  • Code generation or debugging that materially affects results
  • Simulations or numerical analysis
  • Data analysis or statistical analysis

Any substantive use must be disclosed in the paper itself.

Where should I disclose AI use in the paper?

  • Any AI use for figure generation should be mentioned and explained in the figure caption.
  • Any AI use in conducting research should be described where other methods are listed and described.
  • All other uses worth disclosing should be in the Acknowledgment statement.

What should reviewers do when an article appears heavily AI-written?

Contact the editor directly or raise the issue in the confidential comments to the editors of your report. The editor will assess the situation and take appropriate action.

I am not a native English speaker and wish to use AI to improve the English in my manuscript. What advice do you have for me?

You may upload your document to an AI program under the terms described above. Please be aware that AI programs have been reported to introduce errors, including inadvertent plagiarism, and it is the author’s responsibility to check the output for factual and scientific errors.

How can I tell if an AI program will use my input for training data?

The terms and conditions of AI programs are often not forthcoming about this. Look for affirmative statements that input will not be used. If in doubt, choose another program, or forego AI entirely.

A reviewer’s report appears to have been AI-generated. What should I do?

Alert the editor via the notes-to-editor section of your submission. Do not make accusations or assumptions. Criticism in the report may still be valid and should be addressed in good faith. If hallucinated references or scientific inaccuracies make responding impossible, alert the editor before resubmitting.

Can I upload my review to an unrestricted AI tool for light editing and polishing?

No, reviewers should avoid uploading any peer review materials that reveal confidential information.

What should I look for when verifying that an AI tool is acceptable?

  • Formal, legally binding terms regarding data use and privacy
  • An explicit opt-out ensuring user data will not be used for model training
  • Confirmation that all opt-out settings have been correctly applied

Sign In to Your Journals Account

Filter

Filter

Article Lookup

Enter a citation