JCO's AI Policy: ASCO Rules for the Top Clinical Oncology Journal

ASCO's AI policy applies across all ASCO journals:

JCO Clinical Cancer Informatics is particularly relevant, it's the ASCO journal that publishes AI research in oncology. The distinction between research AI (your cancer detection model) and writing AI (ChatGPT editing your manuscript) applies here just as it does at Nature Biotechnology or Nature Methods.

JCO papers directly influence clinical practice. Disclosures should be specific about what AI touched and, critically, what it didn't touch.

For a clinical trial report:

"During the preparation of this manuscript, the authors used ChatGPT (GPT-4, OpenAI, accessed January 2026) to improve the clarity of the Introduction and Discussion sections. No AI tools were used for the statistical analysis (performed by the trial biostatistician, K.L., using SAS 9.4), interpretation of clinical outcomes, assessment of adverse events, or formulation of treatment recommendations. All AI-edited text was reviewed by the principal investigator (A.B.) and the study chair (C.D.) for accuracy against the trial database. The authors take full responsibility for the published content."

For a translational oncology paper:

"The authors used Claude (Claude 3.5, Anthropic, accessed January 2026) to edit the Methods section for language clarity. GitHub Copilot (Microsoft, accessed January 2026) assisted with writing R scripts for the RNA-seq analysis of tumor samples. All code was validated against TCGA reference datasets. No AI tools were used for clinical interpretation or patient outcome analysis. The authors take full responsibility for the content."

For a review article:

JCO forbids using LLMs or AI tools to draft content for Review Articles, so there is no disclosure template to follow: the text must be written by the authors. For a guideline or consensus Special Article, which is not on ASCO's list of forbidden types, ask the editorial office what disclosure it expects before you use any AI tool.

Notice that each example explicitly states what the AI didn't do. At a basic science journal, this level of detail might be unnecessary. At JCO, it is worth doing because:

• Oncologists reading the paper need to know the clinical conclusions are human-generated

• Regulatory agencies may review JCO papers that support drug approvals

• Practice guidelines committees rely on JCO trial reports as primary evidence

• Insurance companies reference JCO publications for coverage decisions

A JCO reader who sees "AI assisted with manuscript preparation" wants to know: did it touch the survival analysis? The response rates? The adverse event tables? The recommended dosing? Answering "no" to these questions in your disclosure builds trust.

ASCO does not publish a schedule of penalties for undisclosed AI use. Its policy says authors remain responsible for the accuracy of all content submitted and are liable for any breach of publication ethics, and its conduct policy says JCO Journals follow COPE algorithms and ICMJE requirements as reference points when author-conduct concerns arise.

The cleanest protection is to disclose in the Methods section before submission, with the tool, version number, date accessed and manufacturer, and to confirm that every co-author knows how AI was used.

JCO and Lancet Oncology are the two journals with the highest direct impact on clinical practice. Both have AI policies that reflect this clinical responsibility. JAMA Oncology (AMA) follows the more prescriptive JAMA Network approach requiring tool name, version, and manufacturer.

Cancer Discovery (AACR) publishes more translational and basic science oncology, where the clinical stakes of AI disclosure are somewhat lower. But the same fundamental rules apply.

For clinical trial reports:

• Never use AI to interpret clinical outcomes, response rates, or survival data

• Don't use AI to draft the adverse events section, this must reflect the actual safety database

• If AI helped with language, be explicit about which sections and confirm clinical content wasn't touched

• Have the trial biostatistician verify that AI didn't influence any analysis code

For practice-relevant papers:

• If your paper includes treatment recommendations, write these yourself. AI-generated treatment advice in JCO would be a serious integrity issue.

• Disclose AI use in the cover letter as well as Methods

For biomarker and translational papers:

• AI can help with bioinformatics code, but validate against established pipelines

• Don't use AI to generate hypotheses about biomarker-treatment interactions in the Discussion

• Patient-level genomic data shouldn't be processed through cloud AI tools

Before submission checklist:

• [ ] AI disclosure in Methods section

• [ ] Explicit statement about what AI didn't touch (clinical data, outcomes, recommendations)

• [ ] Tool name, version number, date accessed, manufacturer, and use case specified

• [ ] Any AI-generated images disclosed (ASCO's AI policy does not address images)

• [ ] Statistical code validated by biostatistician

• [ ] No patient data processed through cloud AI

• [ ] All co-authors (including trial collaborators) aware of AI disclosure

• [ ] Trial sponsor informed of AI use if applicable

A JCO submission readiness check can help verify your JCO submission meets ASCO's requirements before submission.

Comply proactively if:

• You used any AI tool (ChatGPT, Grammarly, Copilot) during manuscript preparation

• The journal requires AI use disclosure in the Methods section

• Your institution has its own AI use policy that may be stricter

Less concerned if:

• You used AI only for grammar/spell checking (most journals exempt this)

• Your use was limited to literature search or reference management