eLife's AI Policy: Funder-Backed Open Access Meets AI Transparency

Since January 2023, eLife operates differently from traditional journals:

No accept/reject decisions. After desk screening, manuscripts that proceed to review are published regardless of reviewer recommendation. The published version includes the full manuscript, peer reviews, the eLife assessment (a summary evaluation), and the authors' response to reviews.

Public peer reviews. Reviewer comments are published alongside the paper. Reviewer identities can be made public if the reviewer opts in.

eLife assessments. Each paper receives a summary assessment with standardized descriptors (e.g., "important," "convincing," "solid") that indicate the editors' view of the work's significance and strength of evidence.

What this means for AI disclosure:

The practical consequence: at eLife, getting caught with inadequate AI disclosure is worse than at most journals because the evidence is public. If a reviewer writes "the Discussion reads as if it was generated by an LLM and no disclosure was provided," that comment will be attached to your paper forever.

eLife now charges publication fees (introduced in 2024), but remains primarily funder-backed by research organizations (Howard Hughes Medical Institute, Wellcome Trust, Max Planck Society, among others). This creates a different incentive structure than gold OA journals:

Lower volume-driven pressure. While eLife now charges publication fees, the journal remains primarily funded by HHMI, Wellcome, and Max Planck. This means publication fees supplement rather than drive revenue, creating less pressure to accept papers quickly compared to fully APC-dependent publishers.

Funder accountability. eLife's funders have their own positions on research integrity and AI. The journal's AI policy reflects not just editorial judgment but the expectations of organizations like HHMI and Wellcome, which have published their own guidance on AI in research.

Lower financial barrier to compliance. eLife's publication fees are moderate compared to many gold OA journals, and fee waivers are available. This reduces the sunk-cost pressure that can discourage authors at more expensive venues from cooperating with disclosure corrections.

eLife's format is relatively flexible compared to STAR Methods journals. Your AI disclosure goes in a standard Methods section:

For a biology paper:

"During preparation of this manuscript, the authors used ChatGPT (GPT-4, OpenAI) to improve the language and clarity of the Discussion section. All suggestions were reviewed by the corresponding author (J.K.) and verified against the experimental results. The authors take full responsibility for the published content."

For a computational biology paper:

"The neural network described in this paper was developed using PyTorch 2.1 (see Methods: Model Architecture). Separately, during manuscript preparation, the authors used Claude (Claude 3.5, Anthropic) to edit the Introduction for clarity. GitHub Copilot (Microsoft) assisted with writing test scripts for the benchmarking suite. All code is available at [GitHub URL]. The authors take full responsibility for the content."

For a paper with extensive analysis code:

"The authors used GitHub Copilot (Microsoft) to assist with writing Python scripts for the single-cell RNA-seq analysis pipeline. All scripts were validated against published reference datasets and are available in the accompanying GitHub repository. ChatGPT (GPT-4, OpenAI) was used to improve the readability of the Results section. The authors take full responsibility for the published content."

The public review consideration

When writing your AI disclosure for eLife, remember that reviewers will comment publicly. A thorough disclosure preempts reviewer concerns. If your disclosure is vague, "AI tools were used", a reviewer might publicly ask for specifics, creating a visible record of inadequate initial compliance. A detailed, upfront disclosure avoids this.

The preprint point is important. eLife encourages preprint posting and accepts submissions via bioRxiv and medRxiv. If you used AI for the preprint version and then submit to eLife, the disclosure should cover AI use during all phases of manuscript preparation, including the preprint stage.

The response to reviewers point is uniquely relevant at eLife. Since reviewer responses are published, using AI to draft your response creates a publicly visible document. eLife doesn't technically require disclosure of AI use in review responses, but transparency is the journal's core value, if reviewers can tell your response was AI-generated, it undermines the dialogue.

During review:

• Reviewers may publicly flag AI concerns in their review
• Editor requests disclosure addition
• The review flagging becomes part of the permanent public record

After publication:

• The paper, reviews, and eLife assessment are already published together
• A post-publication correction adds to the public record
• Expression of concern or retraction for serious cases
• COPE guidelines apply for investigation

The public accountability factor: At most journals, the worst case for non-disclosure is a private conversation with an editor followed by a published correction. At eLife, the entire chain of events is visible: the paper, the reviewers' concerns, the editor's assessment, and any subsequent corrections. This transparency creates a stronger deterrent but also means that legitimate oversights become publicly visible alongside deliberate concealment. The advice is simple: disclose thoroughly upfront. You won't regret over-disclosure, but you'll definitely regret under-disclosure if it's flagged publicly.

eLife is the only journal in this comparison with public peer review and no accept/reject decisions. This combination creates the strongest AI disclosure accountability mechanism of any major journal, not because the rules are stricter, but because the transparency makes non-compliance much more visible.

For all submissions:

• Disclose thoroughly in Methods. Remember, reviewers will comment publicly, and vague disclosure invites pointed questions.
• If you posted a preprint and used AI at any stage, include that in the disclosure.
• Keep in mind that eLife assessments are public. An "incomplete" or "inadequate" assessment partly driven by AI disclosure concerns becomes a permanent quality signal.

For computational biology and bioinformatics:

• eLife publishes significant computational work. Deposit code publicly and note which portions were AI-assisted.
• Benchmarking papers should specify whether AI helped with benchmarking code or parameter selection.

For structural and molecular biology:

• AlphaFold and similar tools are research methods, not writing tools, describe them in Methods as computational approaches.
• If AI helped write analysis scripts for cryo-EM or crystallography, that's a writing disclosure.

For neuroscience and systems biology:

• Don't input participant data into cloud AI tools
• Analysis pipeline code should be independently reproducible without AI tool access

Before submission checklist:

• [ ] AI disclosure in Methods section
• [ ] Tool name, version, and use case specified
• [ ] No AI-generated images
• [ ] Code deposited in public repository
• [ ] Disclosure covers preprint stage if applicable
• [ ] All co-authors aware of disclosure
• [ ] Disclosure is specific enough to preempt reviewer questions

A free manuscript assessment can help verify your eLife submission meets the journal's transparency standards.