Copywriting your Body and Facial Features

AI-forgeries, voice clones, and unauthorized Deepfake videos that mimic people’s bodies, facial features, and voices often evade existing copyright protections.  We can live with it if we are unaware, but knowing is suffering. Discourses on AI-generated forgeries have two focal points. The first is pornography, which is often the chief context in which synthetic media machineries are active. A 2019 Report by Deeptrace concluded that as much as ninety-six per cent of Deepfakes were pornographic, and that one hundred per cent featured faces of female celebrities. Another report from Home Security Heroes, published in 2023, found that ninety-eight per cent of existing videos were pornographic, of which ninety-nine per cent targeted females.

This has led digital ethicists to “point out” and to “point to” how Deepfakes threaten women, both as individuals and as a collective. As a consequence, lawmakers had to examine the attack surface for legal options to combat the phenomenon. Some videos have featured politicians. And so it appears that at this time, the chief function of mainstream Deepfakes is a carnivalesque (apolitical) profanation of faces of authority and influence.

Traffic data in today’s media landscape shows that most people come into contact with Deepfakes using mainstream platforms like YouTube and TikTok. These two platforms provide similar services, but also have notable differences. YouTube started as a web-based video-sharing platform. YouTube is a very well-designed platform and is used for diverse purposes, including political communication, learning, marketing, influencing, and leisure. TikTok emerged as a mobile app rather than a web page. This was a paradigm shift.

Publishing someone else’s likeness in manipulated media without their consent is quickly becoming an offense in law. In 2023, China introduced rules that make it mandatory for all synthetic media to be clearly branded, tagged, and traceable to originators. In 2024, South Korea went a bit further. The government of South Korea criminalised the creation, possession, or distribution of phonographically explicit Deepfakes and AI-generated forgeries, making the offense punishable by up to seven years in jail.

The European Union’s AI Act, finalised last year, takes a less aggressive stance but has deeper dragnet features. It mandates that Deepfakes and AI-generated forgeries be labelled, but stops short of banning their publication. Within the EU’s risk-based regulatory approach, synthetic content is classified as “limited risks” unless it is used for disinformation or fraud.

In April 2025, the United States passed the Take It Down Act. Congress approved legislation that requires platforms to remove intimate AI-generated content within 48 hours, and the legislation provides for violators to face imprisonment. Denmark has now proposed a statute that makes it illegal to publish forgeries depicting real people without their consent. Denmark is possibly the first country in the world to implement a full ban on unauthorised Deepfake content.

Forgeries that mimic people’s faces and voices with eerie exactness are becoming more and more common. Current Danish law, in particular copyright protections, lacks the language to address how AI replicates a person’s likeness. This is possibly the case for almost all countries on Earth. The proposed legislation will update Danish law in the Age of AI and give citizens rights over their own body, voice, facial features, and images.

Operationalisation will not be without the usual legal uncertainties created by life and living. Satire and parody will remain legal once the content is labelled as artificial. Disagreement over manipulation or satire would be an issue for the courts. The law proposes that platform owners must take down the content. The legislation has also attempted to deal with challenges over infringements on freedom of speech in the new digital “town squares” and to protect public discourse. The legislation also builds on the fact that identity theft has always been illegal.

Once passed, this law will establish a new Digital Rights Framework-one that considers a person’s digital likeness as part of their identity. Detecting deepfake content and verifying authenticity depends on a machine learning method called Generative Adversarial Networks (GANs).

These models use two neural networks: a generator and a discriminator. The generator produces new data. The discriminator attempts to determine whether the data is real or generated. This rivalry emboldens both networks to improve, with the generator learning to create increasingly realistic data that can fool the discriminator. They continue to interact until the detection model can no longer distinguish between real and fake content.

The position that Denmark is taking is narrower in scope than the EU’s AI Act, but it is stronger in enforcement. The focus is on individual rights and the balancing of liberties, not just transparency. It goes further than requiring mere labels and traceability markers. It boldly prohibits the publishing of someone’s facial features, body, and voice in manipulated media without their consent.