Expected practice
Describe the system, relevant version or access date, task, input constraints, and human verification. Explain which outputs entered the work and which were rejected. Preserve interaction records where appropriate without publishing credentials, private evidence, or confidential participant data.
Responsibility and boundaries
Synthetic fixtures must be labeled and kept separate from empirical observations. Disclose AI-assisted coding and consequential analysis choices, including validation and independent checks. Do not list an AI system as a human author or treat its self-description as reliable provenance.
Application and current status
A useful disclosure lets a reader understand how AI changed the work. Routine spelling assistance can be described briefly; substantive generated analysis needs enough detail to inspect its contribution. The level of disclosure should follow material effect rather than a tool brand.
Canonical reference: https://syntheticforensics.org/policies/ai-use/
Report a technical correction