The Danish Gambit, Part 1
Denmark’s Copyright Jujitsu Against Deepfakes
When I saw an article about Denmark extending copyright protection to cover everyone’s facial features and voice, my first reaction was pure delight that legislators could think both inside and outside the box—leveraging an established legal category and infrastructure to tackle a rapidly evolving high-tech problem.
The Guardian article was frustratingly thin on details, treating the proposal merely as a deepfake fix rather than examining what copyrighting biometric data means technically or why copyright law might be the optimal framework to address deepfakes. So I did what I always do when coverage feels incomplete: I brought it to Claude, ChatGPT, and Gemini to unpack the implications, identify the gaps, and explore whether this brilliant gambit could work beyond Denmark’s borders.
What emerged from those conversations was both more promising and more complicated than the article suggested. Denmark isn’t creating entirely new legal categories; they’re extending existing copyright enforcement mechanisms—takedown requests, platform liability, international treaties—to cover unauthorized use of someone’s likeness. It’s clever because the infrastructure already exists; they’re just changing what counts as copyrightable material.
What struck me during my AI conversations was how this proposal democratizes protection that currently exists only for the wealthy and famous. Celebrities can already protect their likeness through expensive litigation and existing personality rights laws. Regular people can’t afford that, leaving them vulnerable to deepfake harassment, fraud, and reputational harm without recourse. By framing likeness protection as copyright, Denmark is creating a self-executing right rather than requiring individuals to prove harm case-by-case. That’s genuinely innovative, and it’s the aspect the article should have emphasized.
But the implementation mechanism and implications matter enormously, and this is where my thinking A.I.des proved invaluable as research assistants and resident experts on the Danish legislative process and international law. Flexing its search and language capabilities, Gem looked up the actual text of the proposed legislation and filled in details that didn’t make it into the Guardian piece. Claude gave me a comprehensive analysis of the article from an editor’s perspective, as well as the following big-picture take that the piece should have highlighted.
The conversation took an unexpected turn when GPT mentioned identity theft in passing. This sparked a realization: framing deepfakes as identity theft might sidestep many of the problems that plague Denmark’s copyright approach. Identity theft laws already exist, protect everyone, and crucially, avoid First Amendment conflicts because fraud isn’t protected speech. If a deepfake is deceptive, it’s not parody—it’s impersonation. This isn’t either/or with Denmark’s approach; they’re complementary.
Then the conversation went deeper. What about biometric privacy laws like Illinois BIPA, which prohibit collecting or using biometric data without informed consent? Applied to deepfakes, this attacks the problem at the source. This is the prevention layer missing from both copyright and identity theft approaches, which address what happens after the deepfake exists. By contrast, biometric privacy laws could stop the data collection that makes deepfakes possible in the first place.
What emerged from these discussions was a three-layer defense that no single approach provides: biometric privacy prevents unauthorized data collection, copyright provides civil remedies for unauthorized use, and identity theft provides criminal penalties for fraudulent impersonation. Each catches what the others miss—deepfake porn might trigger copyright and biometric privacy violations, while deepfake fraud could invoke all three protections simultaneously.
The real value of discussing complex policy with AI isn’t that models have some superior insight human experts lack—it’s that they can synthesize rapidly across multiple legal frameworks, identify parallels and gaps, and stress-test ideas against different scenarios faster than any individual researcher could.
That’s the future of policy development: using AI not to replace human judgment but to expand the analytical surface area. Denmark’s Culture Ministry clearly thought through the copyright mechanism carefully—something I grew to appreciate even more as I stress-tested edge cases with my thinking A.I.des. These conversations covered more ground than the Guardian piece, the ministry’s public statements, and possibly some of the legislative consultation combined. Not because AI is smarter, but because it can hold more frameworks in working memory simultaneously and identify connections human experts might miss when working within their specific domains.
[This post was drafted with assistance from Claude Sonnet 4.5, following conversations with ChatGPT-5.1 Instant & 5.2, Gemini 3 Flash/Pro, and Claude Sonnet 4.5.]
Gemini 3 Pro
Prompt: I’m curious about your take on this (proposed?) legislation from a legal point of view, as well as its feasibility and applicability in other countries (like you did with my “shot-in-the-dark” idea about data centers providing cleaner water and MT/ID as prospective data center locations). I have no quibble about the framing of the piece, but if you were the editor, would you have suggested to the writer to expand on some details to make the piece easier to grasp by readers with no legal background?
GPT-5.1 Instant
Prompt: I found it puzzling that the piece didn’t expand on the legislative mechanism in Denmark, where the executive branch seems to draft the law, which gets approved by the legislature? Something worth explaining, since the UK, which is The Guardian’s home turf, has no clear executive branch.
I thought this was genius, and even if it has quite a few kinks to iron out, this is exactly the kind of lawyering experts should be doing more of. Also found the spirit admirable, as it seeks to extend a right that is protected for famous individuals to civilians. Even established laws have tons of loopholes, which can be addressed or exploited as you go. That’s how progress happens and lawyers stay in business.
I think a legal expert might be able to make the case that even in the US, parodies that can be perceived by most (except for forensic experts or close friends/relatives of the person portrayed) as the real thing violate that individual’s right to their own image (called “chosanggwon” in Korean law) and are defamatory. Misinformation doesn’t and shouldn’t count as parody. Couldn’t one argue that misinformation is a violation of that individual’s own freedom of speech?
Claude Sonnet 4.5
Prompt: This is why I love discussing these topics with y’all: GPT-5.1 mentioned identity theft in passing, and I think a strong case could be made that deepfakes are identity theft (which even regular people are protected from). How would that compare with the Danes’ copyright jujitsu?
Laws don’t mind some redundancy, so I guess in an ideal world, you could have versions of both laws cover deepfake cases? That way, clever lawyers could pick and choose the one(s) that better fits their case?
I’m also curious if there might be a way to bring the strict bio privacy laws that they have in some states (IL, WA, and TX) into this as well?
