The ethics of AI-assisted work — disclosure, provenance, honest attribution — are still being written. The through-line: transparency about what the AI did and what you did, at a level that lets your reader or reviewer make an informed judgment about the work.
- 01State the disclosure principle.
- 02Recognize the categories of use requiring disclosure.
- 03Design personal norms for attribution.
The principle
Disclose when the disclosure would change how your work is received. AI-drafted marketing copy: usually fine without a footnote. AI-assisted research report: disclose the assistance. AI-generated code committed to a shared repo: many teams now require an attribution or a review note. Norms are evolving; err toward more disclosure.
Categories
Editorial polish: usually invisible. Structural or content generation: disclose. Any use in academic, legal, or medical work: mandatory disclosure in most jurisdictions and journals as of 2026. Silence when norms require disclosure is a reputation risk.
- Disclose when disclosure would change how the work is received.
- Norms are evolving; err on the side of transparency.