The rapid development of artificial intelligence has transformed the creation of synthetic media, including highly realistic voice replicas generated from relatively small audio samples. As AI voice-cloning tools become more sophisticated, companies, artists, brands, and legal practitioners are increasingly examining how existing intellectual property laws apply to digital identity and unauthorized impersonation.
Recent trademark filings involving voice recordings have attracted attention as rights holders explore additional legal mechanisms to address AI-generated deepfakes and unauthorized commercial uses of voice and likeness. At the same time, these developments have highlighted an important legal reality: trademark law was designed to protect commercial source identification, not personal identity itself.
This distinction is central to ongoing discussions surrounding the proposed federal NO FAKES Act in the United States.
Why Voice-Related Trademark Filings Are Receiving Attention
In April 2026, Reuters reported that Taylor Swift filed trademark applications with the U.S. Patent and Trademark Office (USPTO) covering two audio clips of her voice and one image associated with her public identity. The applications were filed through TAS Rights Management.
According to Reuters, trademark practitioners viewed the filings as a potential mechanism for strengthening protection against unauthorized commercial uses of Swift’s voice and likeness, including AI-generated deepfakes. While the filings themselves do not create broad ownership rights over a person’s voice, they illustrate how trademark law may be incorporated into wider brand-protection strategies.
The filings have also drawn attention to the broader question of whether existing intellectual property frameworks are sufficient to address AI-generated impersonation.
What Trademark Law Protects
Under the U.S. Lanham Act, trademarks function as source identifiers. Their primary purpose is to help consumers distinguish the source of goods or services and to prevent confusion regarding sponsorship, affiliation, or endorsement.
Trademark protection is not limited to words or logos. The USPTO recognizes sound marks, which may qualify for trademark protection when a sound functions as a distinctive indicator of commercial sources.
As a result, a recognizable voice recording or audio clip may potentially serve as a trademark if consumers associate it with a specific source of goods or services.
However, trademark protection remains limited in scope. Trademark law does not create a general ownership right over an individual’s voice, likeness, or identity. Instead, protection depends on the role of the mark in identifying commercial origin and preventing consumer confusion.
This distinction becomes particularly important in AI voice-cloning disputes. Many AI-generated voice outputs do not reproduce an existing copyrighted recording. Rather, they generate new audio that resembles a particular individual. In such situations, trademark claims may depend on factors such as consumer confusion, false endorsement, or related theories under the Lanham Act.
Existing U.S. Protections Remain Fragmented
At present, legal protection against unauthorized voice cloning in the United States is derived from multiple legal frameworks rather than a single comprehensive federal statute.
Depending on the circumstances, claims may arise under:
- State right-of-publicity laws
- Trademark law
- Copyright law
- Unfair competition law
- Consumer protection statutes
- False endorsement provisions under the Lanham Act
One of the principal challenges is that right-of-publicity protections are governed primarily at the state level. States such as California, Tennessee, and New York recognize publicity rights to varying degrees, but the scope, duration, and remedies available differ significantly across jurisdictions.
As a result, legal protections for voice and likeness can vary depending on where a claim is brought.
The NO FAKES Act
The Nurture Originals, Foster Art, and Keep Entertainment Safe (NO FAKES) Act is a legislative proposal intended to establish federal protections for unauthorized digital replicas of an individual’s voice or visual likeness.
The bill was initially introduced in 2024 and subsequently reintroduced with bipartisan support. It has received support from organizations and stakeholders across the entertainment and technology sectors, including SAG-AFTRA and several major technology and media companies.
If enacted, the legislation would create a federal legal framework addressing unauthorized AI-generated digital replicas. Unlike trademark law, the proposed framework focuses directly on the replication of a person’s identifiable voice or likeness rather than on commercial source identification.
Supporters of the legislation argue that a federal framework could provide greater consistency than the current patchwork of state publicity laws and related legal doctrines.
Because the legislation has not yet been enacted, the precise scope and practical application of its provisions would ultimately depend on the final statutory language and subsequent judicial interpretation.
Strategic Implications for Rights Holders
For organizations managing valuable brands, personalities, and media assets, AI-generated impersonation presents both legal and commercial challenges.
As a result, some rights holders are evaluating whether distinctive voice recordings, signature audio assets, and other brand-related elements may qualify for trademark protection.
At the same time, businesses are increasingly reviewing agreements involving:
- Voice rights
- Licensing arrangements
- AI training permissions
- Digital replica provisions
- Talent and endorsement contracts
These efforts reflect a broader trend toward proactive governance of digital identity assets as generative AI technologies become more widely deployed.
International Developments
Other major jurisdictions are also evaluating how existing legal frameworks apply to AI-generated voice and likeness replication.
The European Union’s AI Act focuses primarily on AI governance, risk management, and transparency obligations rather than establishing a standalone right over voice or likeness.
Similarly, the United Kingdom currently relies on a combination of intellectual property, privacy, passing off, and data protection laws rather than a dedicated statutory framework specifically governing AI-generated digital replica.
As policymakers continue to examine the implications of generative AI, legal approaches to voice and likeness protection remain an evolving area across multiple jurisdictions.
Conclusion
Trademark law can provide useful tools for addressing certain forms of AI-generated impersonation, particularly where consumer confusion, false endorsement, or misuse of brand identifiers may be involved.
However, trademark protection is fundamentally designed to protect source identification in commerce rather than to create comprehensive ownership rights over a person’s voice or likeness. The proposed NO FAKES Act seeks to address that distinction by creating a federal framework specifically focused on unauthorized digital replicas of identifiable voices and likenesses.
For legal teams, brand owners, and public figures, the emergence of AI-generated voice cloning underscores the importance of understanding both the capabilities and limitations of existing intellectual property protections as policymakers continue to debate new approaches to digital identity rights.
