Denmark is on the verge of enacting a sweeping law that would outlaw the distribution of non-consensual deepfakes, a move that has ignited serious concerns over expanding government control of online expression. While the stated aim is to shield people from malicious impersonation through AI-generated media, the approach raises red flags about censorship and the boundaries of speech.
Unveiled in April 2025, the legislation would criminalize the use of synthetic media that mimics someone’s voice or appearance without their explicit permission. Under the proposed framework, individuals who discover unauthorized use of their likeness could demand removal, obligating platforms to take the content down once notified.
Though exceptions have been made for satire and parody, these are narrowly defined and require conspicuous labeling to be legally permitted. Such qualifiers risk sidelining creative freedom and place a regulatory filter on how humor or commentary is expressed, especially when AI tools are involved.
Minister for Digital Government Jakob Engel-Schmidt emphasized the enforcement angle, stating, “If an individual finds out that someone has made a deepfake video of them without permission, the law will ensure that the tech giants must take it down again.” His remark underscores a legislative mindset focused less on freedom of expression and more on compelling platform compliance.
This legislative push arrives amid a spike in AI-related deception, including fraudulent voice recordings used in scams and synthetic pornographic material involving public figures. Denmark’s government sees this as justification for preemptive legal intervention. But the risk lies in how this precedent could be expanded or misused. Once a system is in place to decide what is “authorized,” the line between protection and suppression can quickly blur.
While the European Union’s AI Act already mandates the labeling of machine-generated content, it stops short of banning the use of someone’s likeness. Denmark’s law would go significantly further, asserting that biometric resemblance constitutes private property; a notion that, if adopted broadly, could throttle artistic reinterpretation, journalistic reenactments, and anonymous political speech.
The proposed law builds on a prior 2024 Danish regulation limiting deepfakes in political ads. That narrower restriction has now grown into a far-reaching mechanism to police digital identity, applying to the general public and not just to those in positions of power. But with that expansion comes the introduction of an apparatus that, critics fear, could eventually serve more repressive purposes.
Practical enforcement is another unresolved issue. Identifying whether a deepfake was made with or without consent, especially when content is uploaded from outside Danish jurisdiction, poses serious logistical and legal complications. Yet perhaps more pressing is the fundamental question of whether such a law imposes an unacceptable trade-off: sacrificing open discourse in the name of digital security.
