AI disclosure that survives review
Most AI disclosure problems are not caused by using AI. They are caused by discovering, after selection, that nobody can explain which parts were generated or who made the creative decisions.
Keep a production log from day one
Record the tool, version, date, prompt or input, output used and what changed afterwards. A screenshot or saved project is more useful than a memory written after the deadline.
Separate pre-production, production and post-production. Some festivals require generative AI in each stage; others only require a substantive AI role.
Describe the creative decision, not just the tool
A weak statement says “made with AI”. A useful statement says what the human team decided, what the model generated, what was rejected, and how the final edit, performance or sound was controlled.
Name the models only when you can verify the version. If a platform hides the exact version, say so rather than inventing one.
Do not hide hybrid work
If live-action plates, human performances, stock footage or traditional animation are part of the film, disclose them. A hybrid film presented as “100% AI” creates a credibility risk that is worse than the hybrid label.
Check whether the festival requires a minimum AI ratio, a maximum, or no ratio at all. These are different rules.
Prepare evidence before you need it
Keep project files, generation records, source assets and a short one-page AI statement. Some organisers ask for evidence within days of selection or after an award.
Checklist
- Tool, version and date logged
- Human creative decisions described
- Hybrid elements disclosed
- AI ratio rule checked
- Project files and source assets retained
- One-page AI statement ready