Co-authored by Sabrina Alonso and Darsshana Kannan, Summer Law Students

As deepfake technology blurs the line between reality and fabrication, governments are scrambling to respond. Denmark has unveiled one of the boldest proposals yet: a law that would extend copyright protection to a person's likeness, voice and style.

Although Denmark's approach is undoubtedly ambitious, it may create as many problems as it aims to solve.

Generative artificial intelligence (GenAI) is now embedded in daily life, powering everything from social media filters to voice assistants. But alongside its conveniences comes a darker side: deepfakes, hyper-realistic audio, video and images that can put words in someone's mouth or actions in their body. Once a fringe novelty, deepfakes are now tools for non-consensual pornography, political manipulation, financial scams, passing off and viral misinformation.

The power of deepfake technology truly entered the cultural zeitgeist in the spring of 2023, when fans of the artists The Weeknd and Drake were delighted by the release of the recording Heart On My Sleeve. The track had all the hallmarks of The Weeknd's and Drake's respective styles and sounds. Within days of its release, it garnered over 15 million views. Despite the track becoming a success among fans, there was a glaring problem: neither artist was involved in its creation.

Eventually, the recording was attributed to the TikTok account Ghostwriter997, who delighted in their use of GenAI to create the deepfake track. As it reached a fevered pitch in its popularity, the recording suddenly vanished across social media platforms.

Heart On My Sleeve is a rare instance where deepfake technology left a rather benign impact on its subject as it was taken down fairly quickly. However, for everyday people who become victimized by this technology, this is often not their reality.

The copyright blindspot that deepfakes exploit

The Heart On My Sleeve saga highlights a critical gap in copyright law. In response to the growing issue of digital replicas, in 2024 the United States Copyright Office published a report, Copyright and Artificial Intelligence, Part 1: Digital Replicas. In this report, the US Copyright Office acknowledges that a person's voice, likeness and style fall into an intellectual property law grey area.

Copyright law generally protects original expressions of ideas and typically requires that these expressions be fixed in a tangible medium. Some of the works protected by copyright include literary, dramatic, musical and artistic works. A person's face, likeness and voice do not normally fit within these categories.

The challenge in applying copyright protection to protect these features is exacerbated by the idea-expression dichotomy, which explicitly prevents the protection of works where the idea and its expression are so intertwined that safeguarding the expression inadvertently limits public access to the underlying idea. In this instance, granting copyright over an individual's name, image and likeness would make it impossible to separate these features from their expression in different works, effectively giving exclusive rights over the person's actual identity and preventing others from using these elements for legitimate purposes.

Where copyrighted works are fed into an AI platform to generate a replica, copyright law might protect the underlying work itself (such as recorded music or a photograph), but not the broader essence of a person's identity. This means that if a deepfake is merely imitating someone's voice, likeness or style without directly copying a protected work, copyright law offers little recourse.

Even when infringement of a protected work can be established, the remedies would generally be available only to the copyright holder. Artists who do not control the rights to their work would be unable to assert copyright infringement claims, even where their own work is used to create a deepfake of their identity.

Denmark's bold but half-baked plan

Denmark is proposing a radical solution to tackle these gaps given the increased prevalence of deepfakes. The Danish government has announced its plan to broaden copyright protection to include an individual's physical likeness, including their face, body and voice.

The proposed bill would allow individuals to hold copyright ownership over their likeness and provide them with remedies available for copyright infringement including take down of unauthorized uses of their likeness and damages recovery. Another interesting aspect of the proposal is that it would hold GenAI platforms accountable for failing to intervene when unauthorized deepfake content is produced using their technology.

If enacted, the bill would establish the most robust protection of personality rights across Europe. It may surpass the strong stances taken in the American Take it Down Act and South Korea's criminalization of deepfake porn. Although there is certainly a need to close the legislative gap in GenAI regulation, many are wondering if the copyright regime is truly the best-placed tool to tackle the issue of deepfakes.

Where do we go from here?

There is a clear need to close the gaps in personality, privacy and copyright law to protect against the harms of GenAI and deepfakes. Copyright, however, may not be the best suited instrument. It protects the expression of an idea, not the subject of that expression, and aims to balance creators' rights with public access.

Personality rights, by contrast, focus on controlling the use of the subject itself such as a person's likeness, voice or identity. Recasting copyright to encompass one's likeness, voice, and identity as protectable subject matter marks a fundamental shift that challenges its core principles and risks constricting the public domain.

While artists and creators may lean towards pushing for a more aggressive approach toward deepfakes within the copyright regime, it could also come back to bite them. If this new approach were to be widely adopted, artists might have to worry about the legal implications of their work evoking the "style," "sound" or "voice" of those prior artists who influenced them. Nonetheless, it appears that Canada is unlikely to adopt Denmark's approach in the near term as there are currently no plans for a standalone copyright bill that addresses GenAI. Until the law catches up, many artists and creators remain vulnerable as this technology grows more sophisticated and accessible.

Layering on protection while the law catches up

While legislation evolves, artists can take proactive steps to protect both their intellectual property and name, image and likeness (often referred to as "NIL"). These include monitoring their digital footprint, embedding metadata to trace unauthorized uses, registering works for copyright and securing contractual protections.

The Heart On My Sleeve incident offers a practical example. AI is not perfect, and in this case it overtly copied a unique trademarked "producer tag" by the artist "Metro Boomin" who frequently collaborates with Drake. Embedding unique tags or signatures can offer similar protection for other creators while legal reforms catch up.

As deepfake and GenAI technologies transform the creative landscape, vigilance is imperative. Artists concerned about their likeness, voice and work should consider retaining an entertainment lawyer who can assist in safeguarding their intellectual property and digital identity.

Learn more by contacting the author or a member of our Entertainment & Sports Law team.