Introduction to the Updated APS Guidelines

The American Physical Society (APS) has released a significant update to its policy on the use of artificial intelligence tools in manuscript preparation and peer review. Announced on June 17, 2026, the revised guidelines expand the permissible applications of AI while strengthening requirements for disclosure, human accountability, and confidentiality. This development comes as generative AI technologies continue to influence scientific workflows across disciplines.

APS publishes the prestigious Physical Review family of journals, which serve physicists worldwide. The policy applies to all submissions to these outlets and reflects the organization's commitment to maintaining rigorous standards amid rapid technological change.

Background on AI in Academic Publishing

Artificial intelligence tools, including large language models like ChatGPT and specialized research assistants, have become increasingly common in research settings. Authors have used them for language polishing, while reviewers have explored assistance with summarizing literature or checking consistency. However, concerns about originality, accuracy, bias, and confidentiality have prompted publishers to develop explicit rules.

Prior to the update, APS permitted only limited AI assistance, primarily for light editing, condensing text, or polishing language, with mandatory disclosure. The previous framework emphasized that AI could not serve as an author and that human researchers retained full responsibility for content.

Other major publishers have followed similar paths. Many now require transparency around AI use while prohibiting the upload of confidential materials into external systems. The APS update aligns with broader trends in the scholarly communication community, where organizations like COPE and ICMJE have issued guidance on responsible AI integration.

Key Changes in the New Policy

The updated APS policy broadens the scope of allowed AI applications. Authors and reviewers may now employ tools for literature synthesis, data analysis support, scientific reasoning assistance, figure generation, translation, and other substantive contributions, provided human oversight remains central.

Disclosure requirements have been enhanced. Users must specify the name and version of the AI tool, describe how it was directed, explain the nature of its assistance, detail how outputs were verified and validated by humans, and provide any additional context needed for transparency.

Confidentiality protections are stricter. Referees must not upload manuscript content or review materials into external AI tools that could compromise the blind review process. Authors face parallel restrictions when handling sensitive peer review correspondence.

Implications for Authors Submitting to APS Journals

Researchers preparing manuscripts for Physical Review Letters, Physical Review A through E, or other APS titles now have clearer pathways to leverage AI productively. For example, an author might use an AI tool to generate initial drafts of methods sections based on detailed prompts, then extensively revise and verify every element. The policy encourages such hybrid approaches but insists on full human accountability.

Practical steps for compliance include documenting tool usage throughout the research process, preparing detailed disclosure statements for the cover letter or acknowledgments section, and ensuring that any AI-generated figures or analyses undergo rigorous human validation against raw data.

Failure to disclose can lead to manuscript rejection or post-publication corrections. APS emphasizes that the goal is not to restrict innovation but to preserve the integrity of the scientific record.

Photo by Andre William on Unsplash

Guidance for Peer Reviewers

Reviewers play a critical role in upholding standards. The policy explicitly allows AI assistance in tasks such as summarizing complex sections or checking for consistency, but prohibits uploading any part of a confidential manuscript into unauthorized AI platforms.

Reviewers must disclose their own AI usage in reports submitted to editors. This promotes transparency at every stage of the process and helps editors assess the independence of evaluations.

APS provides additional resources on its author and reviewer portals to help users navigate these requirements, including examples of proper disclosure language.

Comparison with Policies from Other Publishers

APS's approach shares common elements with guidelines from Elsevier, Wiley, and Springer Nature, all of which stress disclosure and human responsibility. However, the expanded scope for substantive AI contributions distinguishes the new APS framework from more restrictive earlier versions at some outlets.

Smaller societies and university presses continue to refine their own rules, often referencing APS and similar leaders as models. The variation across publishers underscores the importance of checking specific journal guidelines before submission.

Links to related discussions on evolving publishing norms appear in recent analyses of industry trends.

Impact on Research Integrity and Reproducibility

By requiring detailed disclosure and verification steps, the policy aims to mitigate risks such as hallucinated references, biased outputs, or untraceable data manipulations. Human verification of AI suggestions remains mandatory, reinforcing the principle that technology serves as a tool rather than a replacement for expertise.

Case studies from early adopters of similar policies show that transparent AI use can accelerate literature reviews without compromising quality when paired with careful oversight. APS's framework encourages best practices that could improve reproducibility in physics research.

Stakeholder Perspectives and Reactions

Early feedback from the physics community highlights appreciation for the balanced approach. Many authors welcome the clarity on broader AI applications, while reviewers value the explicit confidentiality safeguards. Editors note that the policy reduces ambiguity during the review process.

Some researchers express caution about over-reliance on AI, emphasizing the irreplaceable value of domain expertise in interpreting results. APS has indicated it will monitor implementation and solicit ongoing input from authors, reviewers, and editors.

Photo by Krists Luhaers on Unsplash

Future Outlook and Ongoing Monitoring

APS has committed to regular reviews of its AI guidance as capabilities evolve. The organization anticipates further refinements, potentially addressing emerging issues such as multimodal AI models or agentic systems capable of autonomous research tasks.

Researchers are advised to stay informed through the official APS journals site and related announcements. Training resources and webinars on responsible AI use in publishing are expected to expand in the coming months.

Actionable Recommendations for Researchers

Physicists and other scholars planning APS submissions should integrate disclosure practices into their standard workflows. Maintain logs of AI interactions, verify all outputs against primary sources, and consult the detailed policy document early in the writing process.

Institutions may consider developing internal workshops on ethical AI use aligned with APS standards. Collaboration with librarians and research integrity offices can further support compliance.

By embracing these guidelines thoughtfully, the community can harness AI's potential while safeguarding the trust that underpins scientific publishing.