Europe has already begun regulating deepfakes. The EU’s AI Act emphasises transparency, risk management and conformity assessments for high-risk AI, but does not create new individual rights against deepfake misuse. It requires providers of systems that generate or manipulate deepfake media to embed machine-readable markings (e.g. watermarks or metadata) flagging content as AI-created, and obliges deployers (platforms or publishers) to “clearly and distinguishably” label any deepfake outputs as AI-generated.
However, the EU AI Act contains no mechanism for removing individual deepfakes once they are online. Some jurisdictions offer their solutions.
In June 2025, Denmark’s Ministry of Culture announced a proposal to amend the Copyright Law to address the misuse of deepfake technology. The new provisions establish three key protections:
- Identity protection – prohibits the distribution of realistic digital depictions of a person’s appearance or voice without consent;
- Artistic performance protection – extends to non-traditional performances such as improvisations and non-verbal acts not typically covered by copyright; and
- Performer protection – specifically safeguards artists from unauthorised digital imitation of their likeness, voice, or performance.
The proposed amendments introduce protection not only for performing artists but for all individuals whose likeness may be digitally replicated. While it does not prohibit the creation of deepfakes, it restricts their public dissemination without the consent of the person portrayed.
The protection applies only to realistic digitally generated imitations, excluding clearly labelled synthetic content or media that is obviously artificial in nature. Also exempted are uses falling within the scope of satire or parody.
The proposed amendments do not establish direct provisions for compensation or criminal penalties. However, they grant individuals and performing artists the legal right to request the removal of unauthorised digital imitations from social media and other online platforms. Importantly, this right is based solely on the unauthorised use of a person’s likeness, without the need to demonstrate any actual harm, allowing individuals to seek removal on that ground alone.
Affected parties may pursue claims for damages and compensation according to the general principles of Danish law. Additionally, under the EU’s Law on Digital Services, if a platform fails to remove illegal content following a formal notification, the service provider may be held financially liable.
The exclusive right to control such digital representations would extend for 50 years after the death of the imitated person or performer.
It is expected that the proposed amendments will be presented to the Danish Parliament in October 2025, and if adopted, will come into force by the end of March 2026. Despite this progress, several practical challenges remain unresolved. Notably, there is no clearly defined threshold for what constitutes a realistic imitation, which may create uncertainty in the law’s application. Additionally, the law’s limited scope to Danish territory could hinder enforcement against content created, hosted, or shared from abroad. Finally, the absence of provisions for sanctions against creators of deepfake content remains an outstanding issue.
Example of a Microjurisdiction: The States of Guernsey’s Image Rights Framework
Denmark’s approach aligns with solutions found in certain jurisdictions, such as the States of Guernsey, which have established specific “image rights” protections within their intellectual property framework. The Guernsey Intellectual Property Office recognises image rights as a distinct category that grants individuals control over the commercial and public use of their likeness and digital representations. This model offers a clear legal basis to prevent unauthorised exploitation of personal identity in digital media, including AI-generated imitations.
Serbia’s Legal Framework on Deepfake Regulation
Serbia currently does not have a cohesive legal regime for regulating deepfake misuse. Existing protections are dispersed across multiple laws. Some of them are:
- Law on Copyright and Related Rights – protects traditional works and performances, but does not extend to deepfakes misuse as such;
- Law on Obligations – enables claims for violation of personal rights (honour, reputation, privacy) but requires proof of actual harm, limiting proactive removal;
- Criminal Law – penalises unauthorised recording or publication of personal data or likeness, but does not address entirely digitally produced creations. Of course, general rules on fraud can be applied if a deepfake was used to deceive a victim with the intent of gaining material gains;
- Advertising Law – prohibits misleading representation or deceptive use of an individual’s image or other personal attribute in advertising, but applies only to commercial contexts and does not cover non-commercial or other digital impersonations;
- Law on Personal Data Protection – governs the processing of personal and biometric data, including automated processing, but does not explicitly address deepfake misuse.
As a result, individuals in Serbia currently lack a direct and efficient legal mechanism to request the removal of deepfake content in most situations they are used in practice, unless they can demonstrate concrete reputational or material damage.
The information in this document does not constitute legal advice on any particular matter and is provided for general informational purposes only.
