Set authorship against real contributions
An author list should not encode status, courtesy or institutional pressure. It should reflect substantial contribution and accountable responsibility. Discuss roles early, record changes and revisit the decision before submission. CRediT makes activities visible but does not replace a journal’s authorship policy. Contributions below that threshold can be acknowledged precisely and with consent.
Separate related outputs without hiding overlap
Several papers from one research programme are not inherently improper. The key questions are whether research question, data, analysis and contribution remain intelligibly distinct, and whether overlap is disclosed. Reused wording, a conference abstract or repeated results need a deliberate check against journal policy and applicable rights.
Label versions, data access and disclosures precisely
A preprint, submitted manuscript, accepted manuscript and version of record are different objects. Clear labels and links stop readers mistaking one for another. A data availability statement should say what is accessible, under which conditions and why a restriction exists. A competing interest does not automatically invalidate research, but concealing one prevents informed assessment.
Protect confidential roles and constrain technical assistance
Peer review gives access to unpublished ideas and data. Do not transfer that material to an external AI service without first establishing that journal rules, confidentiality commitments and the real data flow permit it. An AI system may generate language, but it cannot sustain responsibility, answer for conflicts or meet a human authorship role.
Address errors and problematic material after publication
Correction protects the research record. The suitable mechanism depends on the reach and effect of the problem: a contained error differs from unreliable central findings. When plagiarism is reported, publishers, institutions and affected people should keep their roles distinct. Preserve originals, specify the concern and establish process before public blame. Image editing is likewise judged by purpose, disclosure and its effect on the scientific meaning.