The Right to Your Own Face: Closing the Deepfake Protection Gap

On a Friday in late April 2026, a holding company controlled by one of the most famous people alive walked into the United States trademark system and tried to claim ownership of a few seconds of sound. The applications, filed by Taylor Swift's TAS Rights Management and spotted by the intellectual-property attorney Josh Gerben, sought to register two recordings of Swift's voice speaking the phrases “Hey, it's Taylor Swift” and “Hey, it's Taylor”, along with an image of her on stage holding a guitar. The filings were not about merchandise or branding in any conventional sense. They were a defensive manoeuvre against artificial intelligence, an attempt to build a federal cause of action around the most basic facts of a human being: how she sounds, and what she looks like.

Swift was not the first to try. Four months earlier, the actor Matthew McConaughey had quietly secured eight trademarks through the US Patent and Trademark Office, including a sound mark on the audio of him drawling “Alright, alright, alright” and a clutch of short video clips of himself standing on a porch and sitting in front of a Christmas tree. His lawyers at Yorn Levine had pursued the strategy precisely so that, in a world where deepfakes proliferate faster than any takedown system can manage, their client would have standing to sue in federal court. As one of those attorneys, Jonathan Pollack, put it, “In a world where we're watching everybody scramble, we have a tool now to stop someone in their tracks or take them to federal court.”

This is what the cutting edge of identity protection now looks like: the richest and most recognisable people on the planet contorting trademark law, a body of rules designed to stop you from selling fake handbags, into a makeshift shield for the human self. It works, after a fashion, for them. The deeper and more uncomfortable question is what it tells us about everyone else. Because if the price of protecting your voice and face from machine replication is a trademark portfolio, a federal litigation budget, and a commercial profile valuable enough to demonstrate financial loss, then the overwhelming majority of people have no protection at all.

How a Body Became a Brand by Accident

To understand why a pop star is filing sound marks instead of invoking some clean, dedicated right to her own voice, you have to understand that no such right reliably exists. What exists instead is a patchwork, and the load-bearing piece of that patchwork in the United States is the right of publicity, a doctrine that was never designed for this and has always belonged, in practice, to the famous.

Its origin is almost comically commercial. In 1953, in a dispute between two chewing-gum companies over the right to print baseball players' photographs on cards, the federal judge Jerome Frank coined the phrase “right of publicity” and, in doing so, recognised something new: that a person's persona had an economic value that could be licensed and protected as a kind of property, separate from the older and more dignified right to privacy. The right of publicity, in other words, was born not as a defence of human dignity but as a recognition of human marketability. It protects the commercial value of your identity. The Supreme Court has examined it only once, in the 1977 case of a human cannonball named Hugo Zacchini whose entire act was broadcast on the local news without his consent, and even then the court framed the harm in terms of lost commercial gate receipts rather than any violation of selfhood.

This commercial DNA matters enormously, because it determines who the right actually serves. If the harm the law recognises is the misappropriation of your identity's market value, then the people the law protects are the people whose identities have market value. A film star whose face sells products has a claim. A session musician whose voice is cloned to dodge fraud-detection algorithms has a far weaker one. A nurse in Leeds whose face is grafted onto a pornographic video, or a teenager in Bengaluru whose voice is synthesised to humiliate her, has, under this framework, almost nothing, because the law was built to ask “how much money did you lose?” rather than “what was done to you?”

The right of publicity is also relentlessly local. In the United States it is a matter of state law, recognised in some form in around half the states and varying wildly from one to the next. A claim that succeeds in California may evaporate at the border with Nevada. This fragmentation is exactly why McConaughey's and Swift's lawyers reached for trademark in the first place. Trademark infringement can be litigated in federal court, with nationwide reach and the prospect of statutory damages and injunctions. The right of publicity cannot. So the celebrity strategy is not really about trademark being the right tool. It is about trademark being the only tool with a federal handle on it.

The Borrowed Tool That Does Not Quite Fit

The trouble with using trademark law to protect a human identity is that trademark law was not built to protect human identities, and it strains audibly under the weight.

A trademark exists to identify the commercial source of goods or services and to prevent consumer confusion about who made what. To register the sound of your own voice, you have to convince examiners that the sound functions as a brand, a distinctive identifier of products in commerce, rather than as, well, your voice. Alexandra Roberts, a professor of media law at Northeastern University, has been cautiously sceptical about whether the famous catchphrases at the heart of these filings can clear that bar. Swift's specific sound marks, she has suggested, have some chance of adoption precisely because of the narrow limits of trademark law's reach, but the casual phrases involved, essentially announcements of a new album, do not map neatly onto what trademarks are supposed to do.

Other lawyers are blunter. Matthew Asbell of Lippes Mathias has pointed out that “you don't just own words or recordings of words”, and that any attempt to enforce a trademark against the general timbre of a person's voice would likely fail, calling it “a very narrow right”. A New York court in July 2025 dismissed claims that suggested celebrities could bring trademark actions based on their likeness as a product, a ruling that could undercut the entire approach. And there is an awkward subtext to McConaughey's particular crusade that captures the contradiction of the whole moment: he is an investor in the AI voice company ElevenLabs and has partnered with it to produce a Spanish-language version of his own voice for his newsletter. The objection, in other words, is not to voice cloning as such. It is to voice cloning he does not control and profit from. The right being asserted is, once again, fundamentally a commercial one.

This is the spectacle that the celebrity trademark wave really represents. It is not the arrival of a coherent identity right. It is the sound of the powerful improvising, jamming the levers of an ill-fitting body of law because the law that should exist does not. As the analysis SME Futures published in January 2026 observed, trademark law is being conscripted to perform the function of identity and privacy law in the absence of any dedicated framework. A tool designed to stop counterfeit trainers is being asked to defend the boundaries of the self, and it will do so only for those who can afford to wield it and who have a commercial self worth defending.

The Ninety-Two Per Cent the Cameras Never Find

The clearest way to see the gap is to look at who is actually being harmed, and to compare that population with the population the law currently protects. They are almost mirror images of each other.

In February 2026, India notified its Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Amendment Rules, drafted in late 2025 and brought into force on 20 February 2026, requiring platforms to prominently label AI-generated and synthetic content with visual markers covering at least ten per cent of the surface area and introducing a three-hour takedown deadline for flagged unlawful material, and the Delhi High Court spent much of 2025 issuing protective orders to a procession of celebrities. The spiritual leader Sadhguru secured one in May. The actor Aishwarya Rai Bachchan secured one in September. The journalist Rajat Sharma secured one in November. Each obtained court-ordered takedown regimes against the AI misuse of their likeness, months before any statute made such protection generally available. The personality-rights machinery worked, and it worked for people with names worth protecting.

Set against this is the research documented and amplified across Indian feminist media reporting, including by Feminism in India, which in its November 2025 coverage of gender, AI and digital violence laid out the texture of the harm as it actually falls. The “Make It Real” study by Tattle Civic Tech and the RATI Foundation found that of the women reporting deepfake and AI-manipulated sexual imagery to the RATI helpline, around ninety-two per cent were ordinary women, not public figures. Roughly one in ten calls to the helpline now involved synthetic sexual content. These are not people whose faces sell products. They have no commercial value to lose in the sense the law recognises, which means that under the right of publicity, the dominant Western tool, they have suffered no cognisable injury at all. They have lost something, obviously. The law simply does not have a word for it.

The asymmetry runs right down to the speed of redress. Platforms, the reporting in this space repeatedly notes, tend to move quickly when a celebrity or a brand complains, because a celebrity or a brand can summon lawyers, press attention and the threat of expensive litigation. The ordinary user encounters opacity, delay and indifference. The same image, the same violation, produces a thirty-six-hour court-ordered takedown for the famous and an unanswered support ticket for everyone else. The question that the feminist legal scholarship keeps returning to is the one the whole field has to answer: whose image is worth protecting, who can afford to defend their personality in court, and how do you extend dignity-based protection to people who suffer social rather than economic harm.

This is the heart of the matter. The injury inflicted by AI replication is, for most victims, a dignitary injury, not a commercial one. It is the injury of being made to appear to say or do something you did not, of having the most intimate facts of your embodiment, your face and your voice, seized and operated by a stranger. The law of celebrity cannot see this injury, because it was built to measure a different thing entirely. A framework for ordinary people would have to begin by naming the harm correctly.

What the World Has Tried So Far

The encouraging news is that legislators across several jurisdictions have begun, haltingly and incompatibly, to grope towards something better. The discouraging news is that the legal analysis Harris Sliwoski published in April 2026, written by Elijah Hartman, is exactly right when it argues that the global rules are already here and they do not agree. Its summary of the position is almost a slogan: the rules are already here, they just do not agree with each other yet. One synthetic output, the firm notes, can trigger multiple legal theories in multiple countries at once. For anyone trying to assert a right, that incoherence is itself a barrier.

Consider the range. The European Union's AI Act takes a transparency approach: Article 50, whose obligations come into force on 2 August 2026, requires that deepfakes be disclosed as artificially generated, with carve-outs for artistic and satirical work. This tells you that content is synthetic, but it does nothing to give you a right over your own likeness within that content. The EU's General Data Protection Regulation does more, treating biometric data, including facial images and voice prints processed to identify a person, as a special category requiring explicit consent under Article 9. In principle this is powerful. In practice it is hedged with the requirement that the data be processed specifically to identify an individual, and it lives inside a data-protection regime that ordinary people find slow and forbidding to invoke.

The United States has moved on the narrowest and most urgent front. The TAKE IT DOWN Act, signed in May 2025, criminalises the publication of non-consensual intimate imagery, including AI-generated “digital forgeries”, and requires covered platforms to remove flagged material within forty-eight hours. The notice-and-removal regime came into effect on 19 May 2026, the date by which covered platforms were required to have the process in place, and the Federal Trade Commission has begun enforcing it, with Chairman Andrew Ferguson sending compliance letters to platforms and civil penalties that can reach $53,088 per violation. This is real protection for real people, and it matters. But it is confined to intimate imagery. It does nothing for the cloned voice used in a scam, the face inserted into a fake endorsement, or the likeness scraped to train a model. The proposed federal NO FAKES Act would go further, establishing the first federal right against unauthorised digital replicas of a person's voice or visual likeness, with statutory damages and a post-mortem right lasting up to seventy years. A revised version was reintroduced in the Senate in May 2026 as S. 4591, led by Senator Chris Coons and backed by around thirteen bipartisan co-sponsors, adding a counter-notice procedure to challenge removals and exemptions for libraries, archives and research institutions; the bill remains pending. It has gathered an unusual coalition of support from SAG-AFTRA to Disney to OpenAI. It has also drawn fire from the Foundation for Individual Rights and Expression as a threat to free expression, and from the publicity-rights scholar Jennifer Rothman, who warns that even the revised bill risks people losing control of their own digital selves through the licences they sign. The earlier DEEPFAKES Accountability Act, which would have mandated watermarking and provenance disclosure, stalled and has not been reintroduced in the current Congress.

At the state level, Tennessee's ELVIS Act, effective from July 2024, expanded the state's right of publicity to cover voice explicitly, protecting an individual's voice “regardless of whether the sound contains the actual voice or a simulation”, and, notably, going after the tools: it prohibits distributing software whose primary purpose is the unauthorised production of a particular person's voice or likeness. California's AB 1836 and AB 2602, in force from January 2025, protect performers living and dead against unconsented digital replicas and void the contract clauses that studios use to extract replica rights from actors who lack representation. These are meaningful, but the clue is in the names. The ELVIS Act is for recording artists. The Californian statutes are for performers. They extend the law of celebrity slightly downwards rather than building a floor under everyone.

The Danish Bet on Owning Yourself

The most radical and the most instructive experiment comes from Copenhagen. In June 2025 the Danish government unveiled a proposed amendment to its Copyright Act that would do something none of the other frameworks attempt: give every individual, not merely performers and not merely the famous, a copyright-style entitlement over their own body, facial features and voice. The proposal remains unadopted, but it is advancing: the amended draft bill was notified to the European Commission under the Technical Regulation Information System procedure on 31 October 2025, and parliamentary adoption is expected during 2026, with the protection set to last for fifty years after the death of the person imitated. Denmark also promoted the approach during its presidency of the Council of the European Union in the second half of 2025.

The mechanism is elegant precisely because it borrows the architecture of copyright, a body of law that already knows how to handle takedown notices, platform liability and compensation without requiring proof of commercial loss. Under the Danish proposal, a person whose likeness is reproduced by AI without consent could issue a takedown notice, claim compensation even in the absence of any reputational or financial harm, and trigger liability for platforms that fail to act. Parody and satire are carved out. The protection extends to all persons, which is the entire point. It treats your face not as a brand you have to have built, but as something you own by virtue of having a face.

The Danish bet is worth sitting with, because it answers the central design question more directly than anything else on the table. By decoupling the right from commercial value, it makes the protection universal. The cleaner whose face is deepfaked has exactly the same claim as the film star, because neither of them has to prove the face was worth money. There are objections, of course. Copyright is a clumsy and sometimes dangerous instrument; treating likeness as property has been criticised, including by Rothman in the American context, as a step that can be used against individuals as readily as for them, since property can be signed away. And a national copyright tweak runs straight into the borderless reality that Harris Sliwoski describes, where a deepfake hosted in one jurisdiction harms a person in another under a third country's law. But as a statement of principle, that everyone owns their own features, the Danish proposal is the closest any government has come to the thing the moment actually requires.

The Five Things a Real Framework Would Have to Contain

So strip away the celebrity theatre and the jurisdictional noise, and ask the question the whole situation poses. If the only protection currently available is the law of celebrity, what would a framework of identity rights for ordinary people actually need to contain? The evidence assembled across these jurisdictions points to five non-negotiable components.

The first is a right grounded in personhood rather than commerce. Every existing tool that works for celebrities fails ordinary people at the same point: it asks for proof of market value or financial loss. A framework for everyone must invert this, recognising that the core injury of having your face or voice replicated is dignitary, an assault on your control over your own identity, and is suffered identically by the famous and the unknown. The Danish model and the dignity-based reasoning emerging from Indian courts both point this way. The harm must be defined as the unconsented use of a person's likeness, voice or biometric identity, full stop, with compensation available regardless of whether a single pound was lost. Anything less re-creates the celebrity gate under a new name.

The second is a low-friction, fast removal mechanism that does not depend on lawyers. The reason the TAKE IT DOWN Act and the Danish proposal matter is that they contemplate a notice-and-takedown system an ordinary person can actually operate, with statutory deadlines, here forty-eight hours, and platform liability for non-compliance. The lived asymmetry documented in the deepfake-abuse research is not primarily about what the law says; it is about whether a victim without a legal team can make anything happen. A real framework must make the removal route as accessible as reporting a stolen card, with enforcement, through a body such as the Federal Trade Commission or a national data-protection authority, that does not require the victim to fund the fight.

The third is coverage of the full range of misuse, not just its most lurid corner. Intimate-image laws are essential but partial. A genuine identity right has to reach the cloned voice used to defraud an elderly relative, the synthetic endorsement that puts words in your mouth, the face swapped into political disinformation, and the likeness ingested to train a model. The ELVIS Act's instinct to regulate the tools, the software whose purpose is unauthorised replication, and GDPR's treatment of biometric data as a special category point towards a framework that governs the whole pipeline from training data to output, rather than chasing individual videos after the fact.

The fourth is consent that cannot be quietly stripped away. The danger that Rothman identifies in the property-based approach, and the abuse that California's AB 2602 was written to stop, is the same danger: that a universal right becomes worthless the moment people are made to sign it away as a condition of employment, of using a platform, or of uploading a photograph. A framework must treat identity consent as specific, informed, revocable and resistant to blanket waivers buried in terms of service. A right you surrender by clicking “I agree” is not a right ordinary people can keep.

The fifth is cross-border enforceability, because identity harm does not respect borders and synthetic media least of all. The fragmentation that Harris Sliwoski catalogues, the EU labelling, the American criminal statutes, the Chinese provider rules, the Brazilian electoral resolutions, is not a transitional inconvenience that will resolve itself. It is a structural problem that a single national law, however enlightened, cannot solve alone. A workable framework needs interoperable standards, mutual recognition of takedown orders, and obligations on the global platforms that host this content, so that a person in one country can compel removal of material generated and stored in another. Without this, even the best domestic right becomes a notice shouted into a jurisdiction that cannot hear it.

Why the Celebrities Are the Warning, Not the Model

There is a temptation to read the Swift and McConaughey filings as good news, a sign that the system is responding. It is more accurate to read them as a diagnosis. When the most powerful people in the culture, advised by the most expensive lawyers, conclude that the best available protection for their own faces is to misuse trademark law, the message is not that the system works. The message is that the system has no front door, and they have been reduced to climbing through a side window that happens to have a federal lock on it.

Ordinary people do not have that window. They do not have a sound mark, a litigation fund, or a commercial profile that turns a violation into a quantifiable loss. What they have, increasingly, is the experience of discovering their voice in a scam, their face in a video they never made, or their likeness folded silently into a training set, and then finding that there is no specific law to invoke, no clean right to removal, and no obvious route to compensation. The right of publicity will not help them, because they are not publicity. The trademark gambit will not help them, because they are not brands. The intimate-image statutes will help only if the harm happens to take that particular form.

The components of a better settlement are not mysterious, and they are not even especially novel. They are visible, in pieces, scattered across the very jurisdictions whose incoherence currently frustrates everyone: personhood-based harm from the Danish proposal and the Indian dignity cases, fast statutory takedown from the American intimate-image regime, tool-level and biometric regulation from Tennessee and the GDPR, anti-waiver consent from California, and the unrealised promise of cross-border coordination that nobody has yet built. The task is not to invent these ideas. It is to assemble them into a single right that attaches to a person because they are a person, and not because they are worth money.

The deepest lesson of the celebrity trademark wave is that we have, for seventy years, been protecting the self through the proxy of the brand, and that the proxy has finally broken. A face was always more than a logo. A voice was always more than a jingle. The arrival of machines that can manufacture both, cheaply, at scale, and for anyone, has simply made the inadequacy impossible to ignore. The choice now is whether the right to one's own face and voice remains, as it has always quietly been, a privilege of the famous, or becomes what it should have been all along: a feature of being human, available to the nurse and the teenager and the pensioner on exactly the same terms as the pop star. The law of celebrity will not get us there. It was never meant to. Building something that will is the unfinished work of this decade.

References and Sources

  1. Variety, “Taylor Swift Files to Trademark Her Voice and Likeness, Apparently to Protect Against AI Misuse”, April 2026. https://variety.com/2026/music/news/taylor-swift-trademark-voice-likeness-ai-misuse-1236731401/
  2. Gerben IP, “Taylor Swift Moves to Trademark Her Voice and Image as AI Threats Grow”, April 2026. https://www.gerbenlaw.com/blog/taylor-swift-moves-to-trademark-her-voice-and-image-as-ai-threats-grow/
  3. Northeastern Global News, Cody Mello-Klein, “Is Taylor Swift Trademarking Her Voice Against AI a Good Thing?”, 30 April 2026. https://news.northeastern.edu/2026/04/30/taylor-swift-ai-trademark/
  4. Variety, Todd Spangler, “Can You Trademark Yourself? Inside Matthew McConaughey's Novel Legal Strategy to Fight AI Theft”, 14 May 2026. https://variety.com/2026/digital/news/trademark-yourself-matthew-mcconaughey-legal-strategy-ai-taylor-swift-1236747220/
  5. SME Futures, “Deepfakes, Identity, and the Law: Why trademark is being forced to do identity and privacy's job”, 27 January 2026. https://smefutures.com/deepfakes-identity-and-the-law-why-trademark-is-being-forced-to-do-identity-and-privacys-job/
  6. Harris Sliwoski LLP, Elijah Hartman, “Deepfakes, Voice Cloning, and AI Impersonation: The Global Rules Are Already Here, and They Don't Agree”, 27 April 2026. https://harris-sliwoski.com/blog/deepfakes-voice-cloning-and-ai-impersonation-the-global-rules-are-already-here-and-they-dont-agree/
  7. Feminism in India, FII Team, “Gender, AI And Digital Violence: Mood Of The Month November 2025”, 1 November 2025. https://feminisminindia.com/2025/11/01/gender-ai-and-digital-violence-mood-of-the-month-november-2025/
  8. Media@LSE, “From deepfakes to dignity: what Bollywood's personality rights battle with AI tells us”, 14 October 2025. https://blogs.lse.ac.uk/medialse/2025/10/14/from-deepfakes-to-dignity-what-bollywoods-personality-rights-battle-with-ai-tells-us/
  9. LiveLaw, “Deepfakes And Dignity: The New Battle For Celebrity Rights In India”. https://www.livelaw.in/articles/celebrity-rights-personality-rights-india-deepfake-misuse-legal-framework-article-21-311287
  10. Congress.gov, “H.R.2794 – NO FAKES Act of 2025”. https://www.congress.gov/bill/119th-congress/house-bill/2794/text
  11. The Foundation for Individual Rights and Expression, “The NO FAKES Act is a real threat to free expression”. https://www.fire.org/news/no-fakes-act-real-threat-free-expression
  12. The Regulatory Review, Jennifer Rothman, “Reintroduced No FAKES Act Still Needs Revision”, 18 August 2025. https://www.theregreview.org/2025/08/18/rothman-reintroduced-no-fakes-act-still-needs-revision/
  13. Time, “Danes Could Get Copyright to Their Own Image Under AI Bill”. https://time.com/7298425/ai-deepfakes-denmark-copyright-amendment/
  14. Schjødt, “Owning the Self: Denmark's Copyright Turn Against Deepfakes”. https://schjodt.com/news/owning-the-self-denmarks-copyright-turn-against-deepfakes
  15. Manatt, Phelps & Phillips, “Tennessee's ELVIS Act Expands Publicity Rights for Individuals' Voices & Protects Against AI Misuses”. https://www.manatt.com/insights/newsletters/client-alert/tennessees-elvis-act-expands-publicity-rights-for
  16. Governor of California, “Governor Newsom signs bills to protect digital likeness of performers”, 17 September 2024. https://www.gov.ca.gov/2024/09/17/governor-newsom-signs-bills-to-protect-digital-likeness-of-performers/
  17. EU Artificial Intelligence Act, “Article 50: Transparency Obligations for Providers and Deployers of Certain AI Systems”. https://artificialintelligenceact.eu/article/50/
  18. GDPR-info.eu, “Art. 9 GDPR – Processing of special categories of personal data”. https://gdpr-info.eu/art-9-gdpr/
  19. Proskauer Rose LLP, “Take it Down Act Signed into Law”, 29 May 2025. https://newmedialaw.proskauer.com/2025/05/29/take-it-down-act-signed-into-law-offering-tools-to-fight-non-consensual-intimate-images-and-creating-a-new-image-takedown-mechanism/
  20. Congress.gov, “H.R.5586 – DEEPFAKES Accountability Act”, 118th Congress. https://www.congress.gov/bill/118th-congress/house-bill/5586/text
  21. Right Of Publicity, Jonathan Faber, “A Brief History of Right of Publicity (NIL)”. https://rightofpublicity.com/brief-history-of-rop
  22. Federal Trade Commission, “FTC Begins Enforcing the TAKE IT DOWN Act”, May 2026. https://www.ftc.gov/news-events/news/press-releases/2026/05/ftc-begins-enforcing-take-it-down-act
  23. Law.asia, “India Tightens Rules on Deepfakes and AI-Generated Content”. https://law.asia/india-deepfake-regulations/
  24. Digital Music News, “The 'No Fakes' Act is Back—Can a 2026 Version Pass Congress?”, 20 May 2026. https://www.digitalmusicnews.com/2026/05/20/the-no-fakes-act-2026/
  25. European Parliament Think Tank, “The Danish approach to copyright and deepfakes: A model for the EU?“, 2026. https://www.europarl.europa.eu/thinktank/en/document/EPRS_ATA(2026)782611

Tim Green

Tim Green UK-based Systems Theorist & Independent Technology Writer

Tim explores the intersections of artificial intelligence, decentralised cognition, and posthuman ethics. His work, published at smarterarticles.co.uk, challenges dominant narratives of technological progress while proposing interdisciplinary frameworks for collective intelligence and digital stewardship.

His writing has been featured on Ground News and shared by independent researchers across both academic and technological communities.

ORCID: 0009-0002-0156-9795 Email: tim@smarterarticles.co.uk

Listen to the free weekly SmarterArticles Podcast

Discuss...