Navigating AI for Scientific Figures
Guidance, policies, and FAQs to help publishing scientists and grantseekers use AI with confidence
When to use AI for scientific figures
Generating ideas
AI can be helpful when you’re creating scientific figures to brainstorm ideas or sketch new concepts.
Presenting internally
AI can help you build slides and posters for internal audiences, before you publish in a journal or submit to a funder.
Submitting to certain journals and grants
Some journals and grant funders accept AI-generated figures. Always check policies before submitting.
Journal and grant-funder policies
Policies around the use of AI-generated figures vary widely in the life sciences. Always check your journal or funder's policy before submitting.
Journals often restrict or prohibit scientific figures that are fully AI-generated. If journals allow AI-generated images, they may require use of a dedicated scientific illustration tool like BioRender.
Grant funders like the NIH explicitly prohibit applications that are “substantially developed” by AI. But most don’t have a policy specific to AI-generated figures.
Both publishers and funders regularly require that the use of generative AI be disclosed.
Elsevier
Explicit AI-image policy: AI is allowed for explanatory images and data visualizations derived reproducibly from data, but not for primary research images or graphical abstracts. Elsevier points authors to dedicated scientific illustration tools, naming BioRender. Disclose the tool, version, and use in figure captions and the AI statement.
Springer Nature (including Nature family)
General AI policy: A green/amber/red risk framework replaces the earlier ban on AI-generated images. Only photorealistic deepfakes are prohibited. Schematic and illustrative figures aren’t named, so they fall under permitted tiers with human oversight, verification, and disclosure. Authors remain accountable for accuracy and integrity.
Science/AAAS
Explicit AI-image policy: Generating, refining, or editing illustrative images, diagrams, or figures with AI is not allowed, except with prior editor permission, disclosed at submission. Editing or formatting research data visualizations is allowed and must be declared in the Acknowledgments with model name and version.
Cell Press (including Cell, Molecular Cell)
Explicit AI-image policy: Follows Elsevier’s policy with stricter rules. AI may be used for explanatory images and data-derived visualizations, declared in legends and Methods. It must not create or alter research images or produce cover art (unless integral to the research). General-purpose generative AI image tools may not be used to create graphical abstracts.
PLOS ONE
General AI policy: Report AI use in a dedicated Methods section (or Acknowledgements), naming the tools, how they were used, how outputs were validated, and what they affected. Fabricating or misrepresenting primary data with AI is prohibited. PLOS doesn’t distinguish schematics from data or research images.
Wiley
Explicit AI-image policy: AI is permitted for data visualizations and research illustrations such as workflow diagrams and conceptual frameworks. AI-edited photographs and AI-generated or modified evidential images are not permitted. Disclose tool, version, date, and how it was used in the caption. Wiley cites BioRender in its example disclosure.
BMJ
General AI policy: BMJ favors transparency over prohibition, covering all formats, including diagrams, photographs, and illustrations. Declare what AI was used, why, and how in the contributor section, with fuller Methods detail if used in research. Editors judge suitability; inadequate declaration can lead to rejection or post-publication correction.
JAMA Network (including JAMA, JAMA Oncology, 11 other titles)
Explicit AI-image policy: AI-created or manipulated clinical images and illustrations are not permitted unless part of a fully disclosed research design. AI-generated visualizations of research data are permitted, with the AI or software name and version in the legend. Other AI use goes in the Acknowledgments. Nondisclosure risks rejection.
IEEE
General AI policy: Disclose AI-generated content, expressly including figures and images, in the Acknowledgments, identifying the AI system, the affected sections, and the level of use. Language editing is exempt. AI-simulated data must be labeled as AI-generated. IEEE sets no separate standard for figures.
Policies as published by each major publisher. Individual journals within a publisher family may be stricter. Last updated October 2026.
NIH
General AI policy: Applications, or sections of applications, substantially developed by AI will not be considered the applicant's original ideas.
Wellcome Trust
General AI policy: Generative AI is allowed in applications, but substantive use (idea generation, data interpretation, literature comparison, abstracts) must be declared on the form. Minimal use needs no declaration. AI must not write an application or section without human involvement, or fabricate or alter data, images, or results.
UKRI
General AI policy: Generative AI may be used cautiously in applications. Substantive use should be disclosed; minimal use need not be. AI must not generate an application or section without human involvement, or be used in interviews. Never enter others’ personal data without consent. Misconduct can lead to rejection.
NSF
General AI policy: Proposers are encouraged to indicate in the project description whether and how generative AI was used. Proposers remain responsible for the accuracy and authenticity of their submission. NSF’s definition of research misconduct expressly covers fabrication, falsification, or plagiarism committed through AI-based tools.
CIHR / Tri-Agency (Canada)
General AI policy: Applicants may use generative AI to draft proposal sections. No blanket disclosure is required (removed in 2026). The named applicant is accountable for the application, which must be true, accurate, and complete. Original sources that AI surfaces and that significantly inform the proposal must be acknowledged.
Last updated October 2026.
You're Secure with BioRender
Our AI models are secured through encryption, authentication controls, and continuous vulnerability monitoring, with regular audits to ensure compliance with industry standards. We hold third-party AI providers to these same standards. To learn more about, visit our Trust Center.
Frequently asked questions
Does using an AI tool expose your unpublished research?
Some AI platforms use submitted inputs like prompts, reference images, and uploaded data to improve their models. For researchers working with unpublished findings or clinical data, this is a real concern. BioRender is SOC 2 certified, so your unpublished research stays private.
Who owns the copyright on an AI-generated figure?
Copyright ownership of AI-generated content is unsettled and subject to active litigation. BioRender's AI generates figures from BioRender-owned icons and assets, so the license transfers directly to researchers, removing copyright ambiguity.
What happens if a journal finds AI-manipulated imagery in your published work?
Journals are actively reviewing published work for AI-generated imagery. Figures built in BioRender are accepted by top journals, including Elsevier, Springer, Nature, and Science, giving researchers a clear, compliant path.
Can I use AI-generated content in top scientific journal submissions?
AI-generated content, including figures and images, is restricted or prohibited by most major scientific journals. Always check your journal's editorial policies before submitting.
Can I use AI generated figures in an NIH grant application?
Per NIH's July 2025 Notice, NIH will not consider applications that are either substantially developed by AI, or contain sections substantially developed by AI, to be original ideas of applicants.
Ready to get started?
Start making professional scientific figures today
