Documentary Makers Address Opportunities and Risks of Selling Footage for AI Training
· GlobeWire Daily Editorial Desk · Source: variety.com

Image: Rafa Sales Ross · Photo source · Resized and letterboxed to 16:9; no AI generation
Specialists at the International Documentary Film Festival Amsterdam (IDFA) recently discussed the complexities of licensing documentary material to artificial intelligence training platforms. Industry experts addressed the evolving landscape of AI aggregation, copyright implications, and practical steps for rights holders.
The Evolution of AI and Archival Scraping
During an industry talk titled “Is AI Stealing My Documentary?” Rights Stuff CEO Wendy Bernfield noted that AI systems have historically scraped web content, trailers, and channels for training purposes without early awareness from creators. In recent years, AI aggregators have emerged to combine outputs from multiple models into unified interfaces.
Bernfield stated that out of roughly 15 current aggregators, only a few focus specifically on harvesting archival images. While these platforms often include built-in rights protections, their compensation models tend to be lower compared to riskier alternatives.
Contract Realities and Legal Challenges
Elizabeth Klinck, an archive producer at E Klinck Research, reported receiving weekly inquiries from filmmakers approached by AI entities seeking material. Klinck highlighted potential pitfalls in lengthy contracts, citing an instance where a filmmaker discovered a restrictive five-day limit on an opt-out clause.
Key advisory points discussed for documentary makers include:
- Future-proofing agreements: Ensuring material derived from archives holds up against errors and omissions insurance and copyright clearances.
- Re-clearing third-party rights: Recognizing that archival material often contains embedded rights requiring additional permissions.
- Specific AI clauses: Adding explicit permissions in new archive acquisitions to allow for potential AI usage and future licensing.
Market Shifts and Niche Demand
First Hand Films CEO Esther Van Messel noted that initial AI data requests focused heavily on mainstream, bulk volumes exceeding 5,000 hours. However, Bernfield observed a recent shift toward curated, niche content as AI developers seek to diversify training data to include underrepresented demographics and diverse perspectives.
Additionally, the rising demand for archival content has prompted some public platforms and research libraries to restrict free access to protect resources from continuous scraping.
Verifying Announcements and Practical Implications
Rights holders and filmmakers considering licensing agreements are advised to carefully review contract terms, consult legal professionals regarding third-party rights, and verify all terms directly with original aggregators or platforms before signing agreements.
AI-assisted reporting; source claims are attributed and have not been independently verified.



