The impact of AI on the film industry
Article 2: Spotlight on Belgium
The use of AI in the film industry is a hot topic today and will likely continue to be for some time. One recent example was The Brutalist” about Jewish-Hungarian architect László Tóth. Discussions arose when word came out that the voices of native-English speaking actors were adapted through AI tools during Hungarian-spoken dialogue, with some even calling for a ban from awards such as the Oscars.
Also, the film “Here”, with Tom Hanks and Robin Wright, sparked debate for its use of AI-technology to de-age the actors, making them appear in all stages of life throughout the film. According to some, the so-called licensing and recycling of old footage prevents new talent from appearing on the screen. The AI used to create the de-aging effect is owned by Metaphysic, a company co-founded by Belgian visual effects expert Chris Umé. Three years earlier, Umé created and published a viral deepfake of Tom Cruise on social media platforms such as TikTok and Instagram.
The influence of AI on the film, tv and wider media sector did not go unnoticed by relevant bodies in Belgium. Given its federal structure, no less than three independent agencies are responsible for culture and media in the Flemish, French and German Communities respectively. By organising seminars and information events, the agencies aim to make creators more familiar with both the benefits and the risks of AI. Key topics are the intellectual property rights for the training of AI systems and the potential redundancy of jobs in the future.
From a legislative perspective, the national implementation of EU Directive 2019/790 on the Digital Single Market (DSM Directive can play a key role through its text and data mining provisions. AI systems often rely on such text and data mining activities, as their development and training requires access to vast amounts of text, images, videos and other data. Book XI of the Belgian Code of Economic Law now lays down the relevant exceptions and limitations for text and data mining, meaning that rightsholders can reserve their copyrights over a work and that AI system providers will need to obtain authorisation before carrying out text and data mining activities on such copyright works. Certain exceptions do apply, such as the mining for the purposes of scientific research. The rules of the DSM Directive have been explicitly confirmed by the new EU AI Act (Regulation (EU) 2024/1689), implying that AI-technology as such is not exempted from the legal obligations relating to copyright.
In contrast, new legislative initiatives, governmental declarations or collective bargaining agreements on specific AI use cases in the film sector remain absent. So far, there has been limited movement toward establishing specific rules on the use of AI for voice actors or for creating deepfakes. Most legislative reforms have instead focused on the potentially criminal aspects of using AI to generate media content, such as fake news or revenge porn.
The absence of such legislative initiatives is however not necessarily problematic. As opposed to other certain countries, such as the UK, Belgium has a legal framework regarding portrait rights or more generally personality rights, which are rights that protect all the elements of a person’s identity which make that person unique and recognizable. According to legal doctrine and case law, examples of portrait rights include the right to one’s body, the right to privacy, the right to one’s image, the right to one’s voice, the right to vote and the right to one’s name. An author or owner of a portrait only has right to reproduce it or make it public if he has the permission of the person portrayed or, for 20 years after the latter’s death, the permission of their successors. The use of AI to reproduce voices or images or to make deepfakes of actors (or make these available to the public) without the permission of the actors concerned is therefore already prohibited under the existing legal framework.
To conclude, AI opens up a range of new possibilities in the film and tv sector, but also faces criticism from stakeholders. Certain negative aspects of AI can however potentially be remediated by the existing legal framework on portrait rights and the implementation of the DSM Directive into national law. As the use of AI in film and tv will become more prevalent as use cases develop further, it will have to be awaited whether other national legislative initiatives are nonetheless needed.
