On June 26, 2025, a new war erupted, this time between Denmark and deepfakes. To win it (or at least try), Denmark’s Ministry of Culture presented a pioneering amendment proposal to the copyright regime.
The Anti-Deepfake Law
This new law recognizes that every citizen now holds legal rights over their voice, face, body, and physical similarities — a direct response to the proliferation of deepfakes and the repercussions they can have on reputation, privacy, and even democracy itself.
The type of illegal content ranges from artificially created videos, audio, or images that mimic realities, to AI-replicated artist performances without consent.
The legal text paves the way for immediate removal requests from digital platforms and civil compensation without direct criminal penalties for users, while providing for heavy fines for tech giants that fail to comply with the rules under the EU’s Digital Services Act.
But not everything falls under the new future law: satire and comedy remain allowed. The law aims to protect freedom of expression without restricting artists and comedians. The goal is to balance innovation with the defense of personal integrity and public debate.
What Are Deepfakes — And How Are They Easily Manufactured?
Deepfakes are AI-synthesized content that realistically reproduces voice, face, or body. They use neural networks (GANs) and audio/video productions in a process that generally involves:
- Data Collection — Photos, videos, or recordings of the target person, often collected from social media.
- AI Training — Tools like DeepFaceLab, FaceSwap, First Order Motion Model, or WaveNet train generative adversarial networks to replicate physiognomy or vocal intonation.
- Synthesis — The AI generates a video, for example, where a fictional person’s mouth reproduces specific text, or audio that mimics the original voice.
- Publication — Many of these tools are open-source and accessible to the general public, some with simple graphical interfaces or code available on GitHub.
In minutes, it’s possible to create a deepfake that convinces viewers without great technical expertise.
Examples That Shocked the World
- The deepfake of Joe Biden during the Democratic primaries in New Hampshire, inserted into a false speech. Or even the first encounter with Barack Obama’s speech that left us confused.
- A video of Mette Frederiksen, former Prime Minister of Denmark, in satirical tone that made the debate about AI’s ethical limits even more urgent.
- Deepfakes of figures like Taylor Swift or Pope Francis, which circulated widely on networks as false news or disconcerting situations.
How to Defend Against Deepfakes — Step by Step
- Source Verification: confirm the origin of videos or audio with independent fact-checks and official verifiers.
- Require AI Warnings: platforms should label AI-created content, as provided by the EU’s AI Act.
- Rapid Reporting: use social media reporting mechanisms as soon as possible deepfakes are detected.
- Guaranteed Legal Access: thanks to the new Danish law, those targeted can demand immediate removal and request compensation.
- Technical Tools: use video/audio forensic analysis solutions like Microsoft Video Authenticator, Deepware Scanner, or Sensity AI.
- Digital Education: user training to question suspicious content and not share without verification.
- Social Resilience: cultivate healthy skepticism in society without precipitating judgments.
Human Stupidity as Fuel
The biggest problem with deepfakes isn’t just technology — it’s our tendency to immediately trust images. A convincing face generates automatic empathy and reduces rational criticism. When minimal verification skills don’t exist, a deepfake can influence elections, reputations, or create collective panic in minutes. We know well what happened in recent elections around the world…
While some click first and think later, creators gain time, reach, and impact, and those defending themselves face generalized disbelief that even presents “true facts” to the contrary.
A Legal Revolution and a Warning for Europe
Denmark, with support from 90% of deputies, is preparing to send the proposal to European instances this autumn, with a horizon for adoption in 2026. The hope is that, during the EU presidency, a new standard of right to one’s own digital identity will spread.
It’s an unprecedented milestone: it’s not just about prohibiting AI pornography (as several countries did in 2024), but about giving citizens copyright of their most personal and inalienable features.
Portuguese Legislation Context
Currently, Portugal follows EU regulations regarding deepfakes, primarily through the AI Act and Digital Services Act. In Portugal, there is no legislation literally referring to “revenge porn.” However, victims can exercise their rights in civil proceedings by obtaining compensation according to Article 486 of the Portuguese Civil Code and in criminal proceedings by depriving the perpetrator/s of their liberty.
Content that is either generated or modified with the help of AI – images, audio or video files (for example deepfakes) – need to be clearly labelled as AI generated so that users are aware when they come across such content under the EU AI Act, which Portugal must implement.
Unlike Denmark’s pioneering approach of granting copyright-like protection over personal identity, Portugal currently relies on existing civil and criminal codes alongside EU-wide AI regulations for deepfake-related issues.
Conclusion
Denmark advances with a law that could become the standard for protecting individuals in the digital world. Technology has made it possible to create instant “counter-truths,” but we cannot surrender truth to illusion. Fighting deepfakes requires:
- Modern laws, like Denmark’s
- Responsible platforms
- Detection tools
- Informed society
- Conscious citizens
Without this, we risk living in a world where “seeing is believing” transforms into “seeing is distrusting.” And at that point, increasingly close (if it’s not already contemporary reality), we will have lost everything that truly matters (to us).
