The Danish Gambit, Part 3
Legal Weasel for a Day
Having explored Denmark’s creative legal framework in Part 1 and the practical enforcement mechanisms in Part 2, I wanted to stress-test the whole approach by playing legal weasel—finding every possible angle a defense lawyer might probe to protect deepfakes from regulation. What I discovered is that every defense collapses when you apply it consistently to other contexts, revealing that arguments for deepfakes aren’t principled positions but special pleading for fraud.
The first argument any defense lawyer deploys is the slippery slope: if faces are copyrightable, everything becomes infringement. But Denmark’s law doesn’t prohibit documentation; it prohibits realistic digital reproduction and publication without consent. There’s a clear distinction between capturing someone in the background of your vacation video versus creating a synthetic video where their face says things they never said. A camera records a past event, while a deepfake manufactures a false future. One is a record; the other is a forgery. The slippery slope only exists if you refuse to distinguish between the two.
The transformative use argument comes next: if unauthorized use of likeness per se is violation, then political cartoons, impersonations, biopics, and AI-generated art all become copyright infringement. But this argument misunderstands what makes something transformative. Parody works by signaling that it’s parody—impersonations are funny because everyone knows it’s a comic playing a certain public figure, not the actual individual. If someone deepfaked an individual so realistically that viewers couldn’t tell, that wouldn’t be parody anymore; it would be fraud. Parody requires three elements: commentary on the original, transformative purpose, and clear indication it’s not the real thing. Deepfakes indistinguishable from reality fail the third requirement. If your defense is “viewers should know it’s fake” but you’ve deliberately made it realistic enough that they can’t know, it’s not parody—it’s deception.
Then comes the “copyright requires creative expression” defense: your face is a biological fact, not a creative work. You didn’t design your facial structure; genetics and chance did. If we accept that facts can be copyrighted, this argument goes, then historical events become owned by witnesses and scientific data becomes proprietary to whoever observed it first. But this completely misses what identity protection actually covers. When publishers buy someone’s life story rights, they’re not paying for creative expression—they’re paying for exclusive access to biographical facts and the right to use that person’s name and likeness. The “creative effort” isn’t in living the life; it’s in the value that identity has. Your face has value not because you sculpted it but because it’s uniquely associated with you—your reputation, relationships, social standing. Deepfakes steal identity capital, not artistic expression.
This is where my conversations with my thinking A.I.des produced the most devastating counter-arguments. ChatGPT brought up fingerprints during one discussion, which reminded me that planting someone’s fingerprints at a crime scene is clearly criminal—obstruction of justice, evidence tampering, fraud. Yet fingerprints are biological facts you were born with, requiring no “creative” effort or daily maintenance. If the “biological facts can’t be owned” argument were valid, planting fingerprints would be legal. But it’s not, because we’ve already decided that biometric identity markers can’t be misappropriated even though they’re biological facts you didn’t actively “create.” Deepfakes are just fingerprint fraud with video.
The agribusiness seed patent parallel that emerged from another conversation is equally devastating. Those companies have successfully argued they can patent seeds that are 99.9% natural evolution, claiming tiny modifications justify intellectual property protection. But then corporate lawyers turn around and argue people can’t have rights in their own faces because “those just biological facts.” So an agribusiness giant gets IP protection for 0.1% modification to naturally-evolved seeds, but individuals don’t get protection for their biological features—a dynamic outcome of constant human labor, which are contributed to biological material every second just to maintain those features (stay alive). If 0.1% modification justifies a seed patent, then 100% life-long maintenance surely justifies an identity right. Either human intervention on biological material creates ownership rights or it doesn’t. You can’t have it both ways depending on whether the plaintiff is a corporation or an individual.
The counterfeiting parallel closes the trap completely. Creating fake currency requires sophisticated technical skill, expensive specialized equipment, detailed study of authentic bills, and artistic talent to replicate subtle details. Yet nobody argues counterfeiters should be protected because they’re “artists” or their work is “transformative.” Why? Because the purpose is deception and the harm is systematic—counterfeit money undermines trust in currency. Deepfakes are counterfeit identity, whose harm is permanent. They require a modicum of skill (compared with counterfeiting) and equipment, but the purpose is deception and the harm is systematic—they undermine trust in visual and audio evidence. If we criminalize counterfeit money to protect currency integrity, we should criminalize deepfake identity to protect evidentiary integrity.
Then there’s the edge case that reveals how carefully Denmark actually thought this through: identical twins. What happens if Twin A authorizes a deepfake of themselves, but it’s indistinguishable from Twin B? The legal weasel argument writes itself: Twin A consented to use of their own likeness; the fact that Twin B happens to look identical doesn’t give Twin B veto power over Twin A’s identity expression. But when I asked Gem about this, it found that Danish parliamentary debates actually addressed this scenario. The resolution is context-dependent: Twin A can use their likeness if clearly identified as Twin A with no reasonable confusion possible, but Twin B can block use if a reasonable person couldn’t make the necessary distinction and harm to Twin B’s reputation is likely. It’s essentially a “likelihood of confusion” test borrowed from trademark law—you can use similar marks if consumers can distinguish them, but if confusion causes harm, the other party can block it.
What makes the twin scenario valuable isn’t that it’s common—it’s that it reveals what the law actually protects. Not just biological features, but identity as publicly understood. If content depicting Twin A would be reasonably understood by viewers as depicting Twin B, then Twin A’s consent alone isn’t sufficient defense. The law cares about public perception and resulting harm, not just technical accuracy about whose biological features were captured. This nuance is exactly what gets lost when legal weasels deploy bright-line arguments about consent or biological facts. But crucially, it’s a nuance that my thinking A.I.des caught and helped me understand.
The Danish gambit isn’t perfect—no law is. But the alternatives are worse. Legal weasels can poke at edge cases all they want. At the end of the day, they’re defending the proposition that realistic impersonation designed to deceive should be legally protected. And when you compare that to how we treat identity fraud, counterfeit currency, forged signatures, or appropriation of life story rights, the answer becomes clear: we already prohibit non-consensual misappropriation of identity markers in every other context. Deepfakes aren’t special. They’re just fraud with better technology. Denmark recognized that, worked through the complications, and built a framework other jurisdictions would be wise to learn from.
The simplest argument is one I nearly overlooked: your face and voice are now standard authentication credentials. My phone unlocks with facial recognition. Some apps use voice authentication. These systems exist because biometric markers are supposed to be uniquely yours and impossible to fake. Deepfakes break that assumption completely. If anyone can scrape your social media images and create a realistic synthetic version of your person, they can potentially unlock your phone, access your accounts, and bypass security systems designed to keep others out.
This isn’t hypothetical. The same technology that creates “entertaining” deepfakes will soon be able to generate the biometric credentials to impersonate you for authentication purposes. There’s no principled distinction between the two—they use identical techniques. When legal weasels argue deepfakes are “just creative expression,” ask them: would they call it “art” if someone used their scraped videos to unlock their phone? GPT assures me the current detection mechanism surpasses the generation capabilities, but I’d much rather be safe and prepared rather than feel sorry after the fact about underestimating the “creative” initiative of enterprising criminals. Creating synthetic versions of someone’s biometric identity markers isn’t expression—it’s forging identity documents. We’ve just been slow to recognize it because the technology is new.
[This post was drafted with assistance from Claude Sonnet 4.5 and Gemini 3 Flash and informed by conversations with ChatGPT-5.2, Gemini 3 Flash/Pro, and Sonnet 4.5.]
ChatGPT-5.2
Prompt: 1. I was watching Murder, She Wrote, where a publisher tried to win over Harry (Jerry Orbach) by feigning interest in purchasing the rights to his life story. Isn’t that pretty much what the Danish law is doing for biometrics? Legal weasels might like to argue that there’s been no creative effort in living one’s life or having a distinctive face/voice, but that’d get shot down pretty fast.
2. With artwork ownership, there are different rights, ownership vs. copyright? How does that work? Trying to see if there’s a connection we can draw to deepfakes.
3. Although legal weasels will immediately go for the parody angle, or fair use, I think that’s very weak as well, since for the parody to be truly funny/biting/effective, viewers need to know that it’s not the real thing. That’s the whole point of parodies. I can’t honestly think of a fair use case that would involve producing a life-like likeness of someone. Deepfakes are forgeries (like forging a signature, another thing that came up on a different episode starring JB Fletcher).
4. There was an Instagram account that was taking others’ content and posting composites, claiming it was interpretive art or some such nonsense. I don’t know what happened to that account but I thought it was misappropriation of others’ posts without their consent that the platform should be taking down. Deepfakes are like that, and even more deceptive because you can’t really tell they’re fake.
5. I guess if you allowed deepfakes as interpretive art, then that same lame argument could be made about fake currency, which requires a high level of skill and specialized equipment. But counterfeiting is a crime, and although the Danes are smartly tackling deepfakes with civil law to fast-track things, I think deepfakes deserve consideration as criminal offenses.
Claude Sonnet 4.5
Prompt: I’ve gotten such a delicious detail while talking to GPT (this time 5.2): fingerprints! Even planting fingerprints is a crime, since the intent is clearly to distort facts about the victimized individual’s behavior. People don’t even actively “maintain” their fingerprints like they would their facial features or voice, but fingerprints are part of your identity that belong to you and no one else.
I thought of one “wedge” that some legal weasel might try to exploit and I’m not quite sure the Danish law addresses it or whether all those different analogies apply. What happens if you’re an identical twin and your twin enables or authorizes the deepfake?
Gemini 3 Flash
Prompt: I thought of one “wedge” that some legal weasel might try to exploit and I’m not quite sure the Danish law addresses it or whether all those different analogies apply. What happens if you’re an identical twin and your twin enables or authorizes the deepfake?
