Artificial intelligence in documentaries: what are the legal pitfalls for filmmakers?

More and more documentary filmmakers are using artificial intelligence (AI) to create unique images, voices or storylines, but this innovation carries legal risks. To avoid fines and reputational damage, creators must pay strict attention to portrait rights, avoid copyright infringement and prepare for the strict transparency obligations for deepfakes under the European AI Act. The golden rule for any production is to ensure proper permissions and document every step of your AI creation process.

Protection of image and voice

One of the most critical issues in documentary filmmaking is the use of an individual's image or voice. Within Belgian and European law, the right to image (the portrait right) and the protection of personal data are strongly established.

As a general rule, to use or mimic someone's voice and appearance through AI, you need explicit consent. This falls under the provisions of the General Data Protection Regulation (AVG/GDPR) and the Belgian right to privacy (including the right to image).

There are exceptions for public figures or celebrities when it comes to current news gathering and information reporting. You may report on them without prior permission. However, this exception is not absolute. When you use AI to generate scenes with public figures that deeply intrude into the private sphere, or when you present misleading information or false facts via a ‘deepfake,’ you cross the line. Such misrepresentations of facts can quickly qualify as slander and defamation (criminally sanctionable) and lead to claims for damages.

Copyright risks of AI-generated output

Documentary filmmakers must be vigilant about copyright infringements when they use output from generative AI models in their movies. Generative AI models (such as Midjourney or ChatGPT) are trained on billions of existing works. As a result, the output they generate is not necessarily “copyright safe”.

Integrating AI-generated images, music or text may therefore violate existing intellectual property rights, as enshrined in Book XI of the Belgian Code of Economic Law.

The approach for filmmakers:

  • Double-check your output: Always verify that the generated content is not too similar to an existing copyrighted work.
  • Documentation requirement: Keep a detailed log of what AI tools you used, what ‘prompts’ you entered, and how the end result came about. Should a claim ever follow, a conclusive record is important for your legal defense.

Europe's AI Act: new transparency rules for deepfakes from 2026

The legal context surrounding artificial intelligence is in flux. A major focus for the film and media sector is the European AI Act.

As of Aug. 2, 2026, Article 50 of this regulation goes into effect, which imposes a strict transparency requirement for deepfakes. This means that as a documentary filmmaker, you are legally required to disclose to viewers that certain scenes, voices or images were generated or manipulated with AI. This must be done in a clear and distinguishable way. The only exception to this is when it is abundantly clear to the average viewer that the effect is creative or satirical, although that line is extremely thin in documentaries (where truth-telling is key).

Frequently Asked Questions

Can I use a politician's AI voting clone without permission?
No, not in most cases, especially if you get the politician to say things he or she never said. While there is broader freedom of information around public figures, creating false statements through AI quickly falls under deception or defamation.

Who is liable if an AI tool infringes someone else's copyright?
As a creator or production house, you are primarily responsible for the content you include and publish in your documentary. Even if an AI model created the infringing content, you, as the party exploiting the material, are liable for the infringement.

Does the entry in the credits provide enough transparency for deepfakes?
According to the spirit of the approaching European AI Act, just an unobtrusive mention in the credits will probably not suffice. Transparency must be effective and clear to the viewer at the moment the manipulation comes into view.

Conclusion

The integration of AI into documentaries offers unprecedented creative possibilities, but the legal playing field is complex and unforgiving. From portrait rights and copyright compliance to the new transparency rules from the AI Act, proactive, legal monitoring of your production in Belgium is not a luxury, but an absolute necessity.


Joris Deene

Attorney-partner at Everest Attorneys

Contact

Questions? Need advice?
Contact Attorney Joris Deene.

Phone: 09/280.20.68
E-mail: joris.deene@everest-law.be

Topics