Where do I disclose AI use at eLife?

In the Materials and Methods section. eLife names no other location, and acknowledgments does not satisfy the policy.

Can I list ChatGPT or an AI tool as a co-author?

No. eLife states AI tools "cannot be listed as or serve as co authors of a paper."

Why can't the striking image be AI-generated?

Because eLife publishes striking images under the Creative Commons Attribution Licence, and it cannot grant those reuse rights over AI generated material. The reason is licensing, not research integrity.

Does eLife have a template disclosure sentence?

No. eLife publishes no template. The examples on this page are our suggested wordings, written to cover what the policy asks for.

What happens if I don't disclose AI use?

eLife's AI policy states no specific penalty. Its general integrity policies allow for post publication notices, expressions of concern, and referral to an author's institution where misconduct is alleged.

Is the eLife policy the same as Nature's?

No, and the difference is substantial. Springer Nature journals are "unable to permit" AI generated images for publication apart from narrow exceptions, while eLife permits them with disclosure.

eLife AI Policy 2026: AI-Generated Figures & Disclosure

Name: SciFig
Author: SciFig

Yes — eLife permits AI-generated figures, provided you describe the tool you used in Materials and Methods.

That one sentence resolves the question most authors are actually asking, but it hides a complication worth ten minutes of your time. eLife does not have a single "AI policy." It has three separate rules that land on three different things you might have made with AI, and they do not say the same thing. One of them is an outright prohibition — and almost nobody knows about it, because it sits in a different section of the guidelines from the AI policy itself.

This page reads eLife's own author instructions rather than summarising other summaries, separates what applies to text from what applies to images, and tells you where the disclosure goes.

Does eLife allow AI-generated figures?

Yes, with disclosure. eLife has no blanket prohibition on using generative AI to produce a scientific figure for your manuscript. Its policy treats AI tools as tools — the guidelines put them in the same category as "imaging or statistical software" — and asks that their use be described in the manuscript.

Three conditions come attached, and all three are stated in eLife's own text:

The use must be described in Materials and Methods. Not the acknowledgments, not the cover letter.
The AI tool cannot be an author. It "cannot be listed as or serve as co-authors of a paper."
You remain accountable. Authors "remain accountable for their contributions, responsible for the proper use of AI tools and the reporting of AI tool-output."

The exception — the one hard ban — is the striking image, the promotional image eLife uses to highlight a Reviewed Preprint. That one cannot be AI-generated at all. Details in the three-rules section below.

What eLife's AI policy actually says

eLife's policy lives under the heading "Use of artificial intelligence tools and technology." The operative passage reads:

"Generative artificial intelligence tools such as ChatGPT or AI using large language models to generate article content are at an essential level just tools as, for example, imaging or statistical software. Whilst such AI tools might be used to aid researchers and authors in doing their work, they cannot be listed as or serve as co-authors of a paper, and their use must be clearly described in the Materials and Methods section."

Two things are worth noticing about how this is written.

It is permissive by construction. The framing device is "just tools… as, for example, imaging or statistical software." Nobody discloses that they used ImageJ as a moral confession; they disclose it because methods sections describe methods. eLife is placing AI in that category, not in the category of suspect practice.

It is written about "article content," not specifically about figures. The clause does not carve figures out, and it does not name them. That absence matters, and it is the reason the next section exists.

Writing, data images, and illustrations: three different rules

This is where authors get caught, because the rule that governs your work depends on what you made, and the three rules sit in different parts of eLife's guidelines.

What you produced	Which eLife rule governs it	What it requires
Text / article content written or edited with AI	Use of artificial intelligence tools and technology	Describe the use in Materials and Methods; AI cannot be an author; you stay accountable
Experimental image data — micrographs, autoradiograms, gels, blots	Image acquisition and presentation	No manipulation "that could lead to misinterpretation"; report the tools and techniques used to acquire and prepare images; supply raw unedited gels/blots as source data
Striking image — the promotional image for a Reviewed Preprint	Reviewed Preprints → Striking images	Cannot be AI-generated. Full stop

The striking-image ban, and why it exists

eLife's wording is unambiguous, and the reasoning is the interesting part:

"These files must be available to use under the terms of the Creative Commons Attribution Licence, which means that we cannot include images generated using artificial intelligence."

Read that causal chain again. The ban is not premised on AI images being inaccurate, misleading, or academically improper. It is premised on licensing: eLife publishes striking images under CC-BY, CC-BY requires eLife to be able to grant downstream reuse rights, and the copyright status of AI-generated images is unsettled enough that eLife will not claim it can. This is a rights problem, not an integrity problem — which is why the ban applies to the one image eLife redistributes under an open licence, and not to the figures inside your paper.

The gap nobody addresses: explanatory illustrations

Now the case that matters most to anyone building a pathway diagram, a mechanism schematic, or an experimental-design figure: eLife's guidelines never name this category.

Handled honestly, in three layers:

What eLife states. There is no clause specifically about AI-generated schematics or illustrations. The AI policy speaks to "article content" without excluding figures. The image-integrity policy that does mention preparation tools is written throughout for experimental image data — its examples are exposure times, acquisition parameters, and uncropped gels.
What follows from the two policies read together. An explanatory illustration is article content, and it is an image whose preparation involved a tool. Both clauses point the same direction: say what you used.
What we suggest you do. Disclose it in Materials and Methods, using the wording in the next section. This is our recommendation for staying inside the policy, not a requirement eLife has published — eLife has not addressed this case explicitly.

One practical consequence is worth stating plainly, because it follows from the rules rather than from anyone's opinion: AI-generated experimental data images are effectively impossible at eLife regardless of disclosure. If your figure contains gels or blots, eLife requires the original raw unedited files as source data. A generated image has no raw file to submit. The requirement forecloses the practice without needing a rule that names it.

AI-generated figures at eLife, disclosed correctly

eLife permits them with one line in Materials and Methods. Make the figure, review every element yourself, and disclose the tool.

Explore the Tool

Where the disclosure goes

Materials and Methods. eLife names one location and it is that one.

This is a genuine point of difference between journals, and defaulting to habit will put it in the wrong place. Acknowledgments is the common instinct — and at eLife it is wrong. A cover-letter mention is a good idea for the editor's benefit, but it does not satisfy the policy, because the policy asks for a description in the published manuscript.

There is a second, easily missed obligation on the image side. eLife asks that "authors should provide information within their submission about the tools and techniques used when acquiring and preparing images." If you used a tool to prepare a figure, that sentence already applies to you, independently of the AI policy.

How to write the disclosure statement

eLife does not publish a template sentence, so the following are our suggested wordings, not eLife's official language. They are built to satisfy what the policy asks for: which tool, what it did, and who checked it.

For an AI-generated schematic or explanatory figure:

Figure 3 was generated using [tool name and version] from an author-written description of the pathway, then edited by the authors to correct component labels and interaction directions. All biological content was verified against the cited literature by the authors, who take responsibility for its accuracy.

For AI-assisted language editing:

The authors used [tool name and version] to edit the manuscript for grammar and readability. No content was generated by the tool, and the authors reviewed and approved all resulting text.

For an AI tool used as part of the research itself (image segmentation, classification, structure prediction):

[Tool name and version] was used for [specific task]. Parameters and settings are given in [location]. Outputs were validated by [method].

If the figure was made with SciFig, the same sentence with the placeholder filled in:

Figure 3 was created using SciFig (scifig.ai). The authors reviewed and edited all elements and take responsibility for the accuracy of the content shown.

Three drafting notes, each derived from what eLife's policy asks for:

Name the tool, and its version if it has one. "AI was used" describes nothing and is not a description. A version number helps where the tool publishes one; where it does not, the name and address carry the identification.
Say what it did and what you did. The policy makes you accountable for the output, so a statement that shows your review is doing real work.
Keep it in Materials and Methods even when it feels like an acknowledgment.

What happens if you don't disclose

Honest answer first: eLife's AI policy states no AI-specific penalty. The clause sets out obligations and stops there — no sanction, no stated consequence. Anyone telling you eLife will retract your paper for an undisclosed AI figure is extrapolating.

What eLife does publish is the general integrity machinery that would apply if a problem surfaced:

AI use sits alongside image acquisition and plagiarism as one of the policies "intended to safeguard the integrity of the work," and staff "perform quality and integrity checks on submissions in line with these policies."
If issues are identified after publication, eLife will investigate and "where appropriate publish a post-publication notice to alert readers."
Where misconduct is alleged, eLife reserves the right to suspend the review process, publish an expression of concern, or ask the author's institution or a regulatory body to investigate.

The realistic risk is not a dramatic sanction. It is that a disclosure gap discovered late is handled as an integrity question rather than a formatting one — and the cost of avoiding that is one sentence in Materials and Methods.

How eLife compares with Nature and Science

The differences here are large, and assuming that a policy transfers between journals is the most expensive mistake available in this area.

	eLife	Nature Portfolio / Springer Nature	Science (AAAS)
AI-generated images in the manuscript	Permitted, disclose in Materials and Methods	Not permitted, with narrow exceptions	Not permitted without explicit editor permission
Stated reason	— (permissive)	Unresolved copyright law + integrity	Copyright + industry standards, position may change
Disclosure location	Materials and Methods	In the image field for permitted exceptions	Cover letter and methods or acknowledgments
AI as author	Not permitted	Not permitted	Not permitted
Hard ban specific to one image	Yes — striking image (CC-BY)	—	—

Springer Nature's wording is that while the legal position on AI-generated images "remain[s] broadly unresolved," its journals "are unable to permit its use for publication," subject to three exceptions — contracted agency artwork, manuscripts specifically about AI (case by case), and generative tools "developed with specific sets of underlying scientific data that can be attributed, checked and verified for accuracy." Science states that "AI-generated images and other multimedia are not permitted in the Science journals without explicit permission from the editors."

So a figure that is entirely acceptable at eLife with one sentence in Materials and Methods may be disallowed at Nature and require editorial permission at Science. If you are preparing for possible transfer between journals, prepare for the strictest policy in your target set. The wider cross-publisher picture is in AI figure policies across journals.

The pattern underneath is worth internalising, because it predicts where these policies go next: the restrictions are driven by copyright and attribution, not by doubts about whether AI can draw. eLife's ban attaches to the image it must license openly. Springer Nature cites unresolved copyright law. Springer Nature's own third exception describes the escape route explicitly — tools whose output "can be attributed, checked and verified for accuracy." The question these policies are really asking is whether you can account for the image and stand behind it.

Using SciFig within eLife's policy

That last point determines what a compliant workflow looks like, so here is how SciFig maps onto what eLife asks for.

You edit the figure, so you can vouch for it. SciFig's output is a layered vector canvas, not a flat picture — every label, arrow, and shape is an editable element. That matters for eLife's policy specifically, because the policy makes you accountable for the output. Correcting a mislabelled residue or a backwards arrow yourself is the same act of authorship as fixing it in any vector editor: the accuracy of every element is something you checked, not something you accepted. Reviewing the figure before submission is not optional here — it is the step that makes the disclosure statement above true.

You keep a source file. Edits happen on an SVG canvas you can return to, so "how was this figure prepared" has a real answer rather than a one-shot raster with no history — which is what eLife's request for information about "tools and techniques used when acquiring and preparing images" is asking you to be able to give.

Export meets submission requirements. Editable PPTX, 8K PNG, print-grade PDF, and SVG, which covers eLife's 300 dpi floor and 10 cm minimum width — the full specification table is in eLife formatting requirements.

Two boundaries, stated plainly because this is a compliance page:

Use it for explanatory figures — schematics, pathways, mechanisms, experimental design. Do not use it, or any generative tool, to produce anything presented as experimental evidence. eLife's raw-source-data requirement makes that impossible anyway.
Not for your striking image. eLife's CC-BY ban applies regardless of how much you edited it. Use a photograph or a micrograph from the work itself.

Disclose it in Materials and Methods using the wording above, and a SciFig figure sits inside eLife's policy rather than in tension with it. If you are starting from a written description of the mechanism, SciFig's text-to-figure tool is the entry point for that workflow.

Make a figure that fits eLife's policy

Editable vector output you verify yourself, plus a disclosure sentence you can paste straight into Materials and Methods.

Try Free

Where to go next

Disclosure is one requirement among several. The rest of this eLife guide:

"What must the figure itself meet?" — 300 dpi, 10 cm minimum width, no cap on figures or supplements. eLife formatting requirements.
"How does the submission work?" — preprint first, $3,750 charged at review rather than on acceptance. eLife submission guide.
"How selective is it?" — 35% of 2025 submissions were sent for review. eLife acceptance rate.
"What about the impact factor?" — there is none, and the reason is Web of Science partial indexing, not the journal's standing. eLife impact factor.

Frequently asked questions

Sources

eLife author instructions — open the Publishing Policies tab for "Use of artificial intelligence tools and technology," "Image acquisition and presentation," and the integrity and misconduct policies; open the Reviewed Preprints tab for the striking-image requirements
Nature Portfolio AI editorial policy — Springer Nature's position on generative AI images and its exceptions
Science journals editorial policies — the "Artificial intelligence (AI)" section, covering AI-generated images and disclosure locations
About eLife — organisation and publish-review-curate model

Policy version. eLife's author instructions carry no version number or "last updated" line, so there is no way to tell from the page itself when the AI clause last changed. Policies in this area move quickly — treat eLife's own page as the current text and check it before you submit.

A note on navigating the source. eLife's author instructions are a single page with tabbed sections, and the tabs are not linkable — a URL ending in #policies will not open the policies tab. Open the page and click Publishing Policies to reach the AI clause.

Boundaries. Every policy quotation above comes from eLife's own author instructions. Two limits are worth stating plainly. We cannot confirm whether eLife's submission forms include an AI declaration field separate from the Materials and Methods requirement, so treat Materials and Methods as the obligation and any form field as a bonus. And eLife has published no clause specifically addressing AI-generated explanatory illustrations — where this page recommends a course of action for that case, it is labelled as our recommendation rather than eLife policy. Journal policies in this area change quickly and without notice; confirm current requirements on eLife's own pages before submitting. See our editorial standards for how these guides are researched and corrected.