Can You Use AI to Make a Graphical Abstract? Journal Policies (2026)

The short answer: it depends on the journal

Yes, you can use AI to make a graphical abstract — the tools exist and they work. Whether you can submit an AI-generated graphical abstract is a different question, and the answer varies sharply by publisher. As of 2026:

• Several major publishers prohibit AI-generated figures entirely — including Nature, Cell Press, Elsevier, Wiley, and Frontiers.
• A few allow them, provided you disclose the AI use — including ACS and PLOS ONE.
• One doesn't use graphical abstracts at all — Science (AAAS).

So before you generate anything, check your target journal's policy. The table below summarizes where the major publishers stand.

Publisher AI-image policies for graphical abstracts (2026)

Publisher	AI-generated graphical abstract?	What their policy says
Elsevier	❌ Not allowed	General-purpose generative-AI image tools must not be used to create graphical abstracts.
Cell Press (Elsevier)	❌ Not allowed	Follows Elsevier's policy.
Springer Nature / Nature	❌ Not allowed	Does not allow generative-AI images, including scientific diagrams and illustrations. Narrow exceptions (licensed stock, tools built on traceable scientific data, AI-about-AI) require disclosure.
Wiley	❌ Not allowed	Prohibits AI-generated images — including AI images edited afterward — and this extends to graphical abstracts; any AI use must be disclosed.
Frontiers	❌ Not allowed	AI-created figures, illustrations, or visualizations are not permitted; recommends tools like BioRender or Mind the Graph; disclosure required.
ACS	🟡 Allowed, with disclosure	Permitted if you disclose the tool and how it was used, and confirm the tool's terms allow commercial use and don't claim ownership of the output.
PLOS ONE	🟡 Allowed, with disclosure	Does not specifically prohibit AI-generated figures, but requires disclosure of AI use and holds you responsible for accuracy.
Science (AAAS)	⚠️ N/A	The flagship journal doesn't use graphical abstracts — it uses a short structured summary instead.

Policies change, and individual journals within a publisher can differ. Always confirm against the official author guidelines for your specific target journal before submitting. (Sources: each publisher's editorial / AI policy pages, accessed 2026.)

Why so many publishers restrict AI images

The concern is scientific integrity. A graphical abstract represents your findings, and a general-purpose image model can invent structures, mislabel components, or fabricate visual detail that looks plausible but is not in your data. Publishers that prohibit AI images are drawing a line around figures that could misrepresent results. Publishers that allow it with disclosure are betting that transparency plus author accountability is enough.

What "disclosure" actually means

Where AI is allowed, disclosure is not optional. In practice it means:

• Name the tool and how you used it — usually in the figure caption, the methods, or the acknowledgements (check where your journal wants it).
• Take responsibility for accuracy — you are accountable for every element in the figure, AI-generated or not.
• Check the tool's terms — ACS specifically requires that the tool's terms allow commercial use and do not claim ownership of the output. Many general image models are acceptable on this point; verify for the one you use, and note that some embed an invisible watermark in their output.

How to use AI responsibly for a graphical abstract

Whatever your target journal's policy, AI is genuinely useful for the parts before a final submission:

1. Drafting. Turning a blank page into a first-pass composition — from your abstract text or a rough sketch — is where AI saves the most time.
2. Sizing and format. Getting the dimensions, DPI, and file format right for your target journal is mechanical work that tooling can handle regardless of image policy.
3. The final figure should then comply with your journal's rules: submit the AI figure with disclosure where that is allowed (ACS, PLOS ONE), or produce the submitted artwork in a permitted way where AI images are prohibited (Nature, Cell, Elsevier, Wiley, Frontiers).

This is how Abfig is built to be used: it drafts a graphical abstract with AI from your abstract or sketch and sizes it to your target journal at 300 DPI, and it includes a DPI checker to verify your final artwork against journal specs. For ACS or PLOS ONE, you can generate and submit with disclosure; for the prohibiting publishers, use it to plan and size, then finalize the figure in a compliant way.

The bottom line

Can you use AI to make a graphical abstract? Technically, always. Should you submit one? Only where your journal allows it, and only with proper disclosure. Match the tool to the policy: ACS and PLOS ONE let you generate and submit (disclosed); Nature, Cell, Elsevier, Wiley, and Frontiers do not — for those, treat AI as a drafting and sizing aid, not the final artwork.

Frequently asked questions

Can I use ChatGPT or Midjourney to make a graphical abstract? You can generate one, but you can only submit it where the journal permits AI images — currently ACS and PLOS ONE among major publishers, with disclosure. Nature, Cell, Elsevier, Wiley, and Frontiers prohibit AI-generated figures.

Do I have to tell the journal I used AI? Yes, wherever AI is permitted. Both ACS and PLOS ONE require disclosure of AI use, and you remain responsible for the figure's accuracy.

Which journals allow AI-generated graphical abstracts? As of 2026, ACS and PLOS ONE allow them with disclosure. Most other major publishers — Nature, Cell Press, Elsevier, Wiley, and Frontiers — do not.

Can I submit an AI-generated graphical abstract as-is? No. Treat any AI-generated figure as a draft: check it for scientific accuracy and confirm it against your journal's format rules and AI policy before submitting.

Related guides

• How to make a graphical abstract
• Graphical abstract examples that got published
• Browse journal specifications

Draft and size a graphical abstract with Abfig →