Generative artificial intelligence tools can assist scholarly work, but they cannot take responsibility for it. The Journal of Mathematics requires transparency about their use from authors, referees and editors alike.

For authors

• AI tools cannot be listed as authors or co-authors. Authorship requires accountability, which software cannot hold.
• Any use of generative AI in preparing a manuscript — for language editing, code generation, symbolic computation, literature searching or drafting — must be declared in a short statement before the reference list, naming the tool, its version and the purpose.
• Routine tools such as spelling checkers, reference managers and standard computer algebra systems used for verification do not require declaration.
• Authors remain fully responsible for the correctness of every statement, proof and computation in the paper, including any produced with AI assistance. AI-generated proofs must be verified independently.
• AI-generated or AI-altered images and figures are not permitted unless the image is itself the subject of the research and this is clearly stated.
• Fabricated references, a common failure of language models, are treated as a serious integrity breach.

For referees

• Manuscripts under review are confidential. Uploading any part of a manuscript to a public or third-party AI service is a breach of that confidentiality and is prohibited.
• Referee reports must represent the referee's own scholarly judgement. Reports generated substantially by AI are not acceptable.
• Limited use of AI to improve the wording of a report the referee has already written is permitted, provided no manuscript content is entered into the tool.

For editors

• Editorial decisions are made by people. AI must not be used to decide the fate of a manuscript.
• Editors must not enter manuscripts, referee reports or author correspondence into external AI systems.
• AI may support administrative tasks such as similarity screening or reviewer discovery, always under human oversight.

Non-compliance

Undeclared use of AI that materially affects the content of a paper is handled under the journal's publication ethics procedures and may lead to rejection or, after publication, to correction or retraction.