“Fake Drake” and the Real Harm: Canadian Privacy Torts Fail to Protect Voice-Cloned Artists

by Ambreena Ladhani, winner of the 2025 Privacy and Access Law Section Student Essay Contest

Introduction

Voice cloning has exploded in the music industry. Historically used to replicate background beats and production tools, artificial intelligence (“AI”) now enables users to “borrow” the vocal signatures of established artists and repurpose them into new music.¹ Many are incentivized to do so, “in order to garner more streams and views.”² And there is no barrier to entry. Right now, anyone can clone the voices of famous artists like Drake, Taylor Swift, and Jay-Z.³ In almost all cases, artists have neither authorized nor consented to the use of their voiceprint for cloning purposes. Yet, their distinctive vocal signatures continue to be repurposed into new, unauthorized, musical works.

Ghostwriter977 used voice cloning technology to create an AI-generated song that mimicked the voices of Drake and The Weeknd.⁴ The song went viral; it quickly amassed 9 million views and was briefly available on TikTok, YouTube, and Spotify before removal.⁵ That is not all. Imagine Drake’s surprise when he discovered that “Heart on My Sleeve” was eligible for two Grammy Awards: Best Rap Song, and Song of the Year.⁶ How? Recording Academy CEO, Harvey Mason Jr., contends that “[Heart on My Sleeve] is eligible, because it was written by a human.”⁷ The lyrics, indeed, may have been written by a human – Ghostwriter977 – but it would be fallacious to claim that the song went viral due to its lyrics alone. The song went viral because it sounded like Drake.

Intuitively, something about “fake Drake” just feels wrong. Celebrities and public figures are more likely to be victims of a voice-cloning attack.⁸ But for music artists, the harm is distinct and profound – voice cloning dilutes the very essence of an artist’s career. The harm cannot be overstated. Certainly, this conduct must be actionable. Drake should have recourse for the “fake Drake” song, and the public seems to agree. Reddit users have noted that “the way [AI] is bleeding into the music industry is alarming.”⁹

But without federal legislation governing the use of AI in Canada, recourse for voice cloning remains uncertain. While the commercial use of an artist’s voice may give rise to liability, it is difficult to predict how existing torts may apply in the context of non-commercial voice cloning. And without adequate protection, artists may be discouraged from creating new music altogether. Indeed, we only understand the value of water when the well runs dry. Though, there is reason for optimism. Canadian courts have acknowledged that “the use of new technologies has created…new possibilities for privacy breaches that require adequate legal protection.”¹⁰ And adequate legal protection is required now, more than ever.

The following research examines the inadequacy of Canada’s privacy tort regime in protecting artists against unauthorized voice cloning, particularly when such cloning occurs without commercial exploitation.¹¹ Part I analyzes Canada’s four recognized privacy torts, and examines the shortcomings of each tort, with a focus on the misappropriation of personality tort as a potential, but flawed remedy. Part II proposes an amended misappropriation of personality tort, expanding its scope beyond commercial appropriation to ensure the tort is actionable even in non-commercial contexts. Part III addresses potential freedom of expression concerns and argues that the proposed framework strikes an appropriate balance between artist protection and freedom of expression. Finally, Part IV offers concluding remarks and acknowledges the limitations of the proposed amendment.

I. The Privacy Tort Regime is Unfit for Voice Cloning

Canada’s privacy tort regime is fragmented, to say the least, comprised of an inconsistent patchwork of legal standards. In Ontario and Alberta, privacy torts are governed by the common law. In contrast, in Manitoba, British Columbia, Saskatchewan, and Newfoundland and Labrador, provincial privacy statutes provide the basis for a broad cause of action: statutory invasion of privacy.¹²

The hybrid statutory-common law governance of privacy torts is already frustrating; to exacerbate the inconsistencies, the standard to find a privacy breach varies between provinces. In Manitoba, the cause of action is established when one “substantially, unreasonably, and without claim of right, violates the privacy of another person.”¹³ In British Columbia, Saskatchewan, and Newfoundland and Labrador, the cause of action requires a higher standard – one must “wilfully and without claim of right, violate the privacy of another person”.¹⁴

At common law, Canadian privacy torts are derived from Professor Prosser’s influential four-tort catalogue.¹⁵ Prosser articulated four distinct privacy torts: (1) intrusion upon seclusion, (2) public disclosure of private facts, (3) false light, and (4) misappropriation of personality. While each of these torts has been recognized in at least one Canadian province, they all present significant limitations when applied to modern privacy challenges, particularly in protecting artists from voice cloning technologies.¹⁶

1. Intrusion upon seclusion concerns “private” affairs

The tort of intrusion upon seclusion was first recognized by the Ontario Court of Appeal in Jones v. Tsige.¹⁷ The tort has not been formally recognized by any other province. And it is unlikely to provide artists with any meaningful recourse for artists who have fallen victim to voice cloning. To succeed in an action for intrusion upon seclusion: (i) the defendant’s conduct must be intentional or reckless; (ii) the defendant must have invaded the plaintiff’s private affairs or concerns without lawful justification; and (iii) a reasonable person would regard the invasion as highly offensive, resulting in distress, humiliation, or anguish to the plaintiff.¹⁸

While it would be relatively easy to demonstrate the first element, that the voice cloning was done intentionally, that action is unlikely to succeed based on the second requirement: the defendant must have invaded the plaintiff’s private affairs.¹⁹ In Caplan v. Atas, the Ontario Superior Court denied the plaintiff’s intrusion upon seclusion claim because the plaintiff’s intruded affairs were not considered “private”.²⁰ In that case, the court reasoned that there was no “violation of the privacy of the plaintiffs” because “these pictures were placed on the internet.”²¹ And it is unlikely that the replication of an artist’s voice, which too, is widely available on the internet, would be considered an intrusion of a “private” affair. Even if satisfied (which is improbable at best), the tort would only provide a viable cause of action in one province – Ontario. The tort is of limited use as a result.

2. Public disclosure of private facts requires “truth in the matter”

The tort of public disclosure of private facts has been recognized more broadly than intrusion upon seclusion. It has been affirmed in Ontario, Alberta, Saskatchewan, and Nova Scotia.²² To succeed in an action for intrusion upon seclusion: (i) the defendant must have given publicity to a matter concerning the private life of another, and (ii) the matter publicized or the act of publication would be highly offensive to a reasonable person and is not of legitimate concern to the public.²³

The same difficulty arises in both intrusion upon seclusion, and public disclosure of private facts – the defendant must have given publicity to a matter that concerns the private life of another. Again, it may be difficult to argue that the publicized AI-generated music concerned the private life of the artist. An artist’s voice is not private. But another significant roadblock halts this tort. The publicized matter must be true.²⁴ And, voice-cloned music is, by definition, false. The very essence of voice replication is rooted in its falsity. And so, this tort must fail too.

3. The tort of false light is helpful but not well-recognized

There is a beacon of light – or “false light,” rather. The tort of false light protects a distinct privacy interest: the right to control the way one is presented to the world.²⁵ False light will be satisfied if: (i) the defendant gave publicity to a manner concerning another person that places the other in a false light, (ii) the false light in which the other was placed “would be highly offensive to a reasonable person,” and (iii) the defendant “had knowledge of or acted in reckless disregard as to the falsity of the publicized matter and the false light in which the other would be placed.”²⁶

For clarity, the plaintiff does not need to show that the misrepresentation is disparaging. Indeed, it is “enough for the plaintiff to show that a reasonable person would find it highly offensive to be publicly misrepresented as they have been.”²⁷ Accordingly, false light is not concerned with reputational harm. Rather, the tort recognizes that there is a distinct harm in “publicly representing someone, not as worse than they are, but as other than they are.”²⁸ Accordingly, “false light can respond even when defamation cannot.”²⁹

Indeed, this is why false light “most adeptly captures the harms” resulting from being impersonated in a deepfake image or video.³⁰ And too, it would appear to provide recourse for the unauthorized cloning of an artist’s voice. A publicly released voice-cloned song would cast the artist in a false light, which a reasonable person would find highly offensive – not necessarily because of the content or lyrics of the song, but due to the fundamental misrepresentation of the artist’s identity.³¹ It would be easy to show that the defendant had knowledge of, or acted in reckless disregard, when creating the new song, as the very act of synthetically replicating someone’s voice implies knowledge of its falsity.

Perhaps, the tort would be satisfied, and thus, may provide recourse for an artist who has fallen victim to voice cloning. But again, there is no case law interpreting false light in the context of digital replications at all. While false light appears to be an appropriate mechanism, it remains a novel tort in Canada. Certainly, it has only been recognized in Ontario, and to date, no other province has followed suit.³² And the Manitoba Court of Queen’s Bench has historically rejected the tort of false light as a viable cause of action under Manitoba’s privacy legislation.³³ While some call for broader recognition of the tort,³⁴ the current lack of recognition introduces uncertainty. This tort may be difficult to pursue as a result. And so, its application remains limited.

4. Misappropriation of personality tort requires commercial gain

Finally, turn to the misappropriation of personality tort, which appears to neatly address the unauthorized replication of personal characteristics. While the tort is rooted in privacy principles, the tort is “proprietary in nature”.³⁵ The tort recognizes that celebrities (and in some cases, other professionals) have a commercial interest in their identity, image, and persona.³⁶ To establish the misappropriation of personality tort, it must be shown that: (i) the defendant misappropriated the name, image, or likeness of the plaintiff; (ii) for commercial gain; and (iii) without consent.³⁷

The misappropriation of personality tort is well accepted in American jurisprudence, otherwise referred to as the “right of publicity.” The American right was first articulated in Haelan Laboratories Inc. v. Topps Chewing Gum Inc., described as the “right of an individual, especially a public figure or a celebrity, to control the commercial value and exploitation of his name and picture or likeness and to prevent others from unfairly appropriating this value for their commercial benefit.”³⁸

Scope of protection is unclear

The common law tort of misappropriation of personality is well-accepted. Different iterations of the tort have also been codified in provinces with privacy statutes under a broader “invasion of privacy” cause of action.³⁹ Perhaps it is reassuring that, in Manitoba, Saskatchewan, and Newfoundland and Labrador, the tort expressly protects one’s voice.⁴⁰ Conversely, British Columbia only protects against the unauthorized use of “the name or portrait of another” when used for “advertisement or promoting the sale of…property or services”,⁴¹ but the statute does not expressly refer to one’s voice.

At common law, indeed the tort protects the commercial image one has in their own “persona.” And while an individual’s persona clearly has value, “such value is not easily defined.”⁴² In Krouse v. Chrysler Canada Ltd. et al, the first case recognizing the tort at common law, the Ontario Court of Appeal confirmed that the tort protects against the appropriation “of another’s likeness, voice or personality.”⁴³ Following Krouse, the common law tort has since been described by courts as “a proprietary interest in the exclusive marketing for gain of his personality, image and name”.⁴⁴ The protection of one’s voice is not always articulated in the common law definition, but given the articulation in Krouse, there is at least an authoritative statement in Ontario that the tort protects one’s voice.

Canadian courts have yet to contend with the misappropriation of one’s voice. Perhaps unsurprisingly, American jurisprudence has indeed held that the right of publicity (equivalent to the Canadian tort of misappropriation of personality) extends to the protection of one’s voice – or a “mimic” thereof. In Midler v. Ford Motor Co., famous musician Bette Midler sued Ford Motor Company for using a soundalike in one of its Ford commercials.⁴⁵ Midler successfully argued that her voice had been appropriated for commercial gain and recovered damages under the right of publicity—the U.S. equivalent of the tort of misappropriation of personality. The Ninth Circuit, in that case, recognized that “[t]he human voice is one of the most palpable ways” that identity is manifested, and an individual’s voice is “as distinctive and personal as a face.”⁴⁶

Interpretations of common law privacy torts in America often influence Canadian treatment, and thus, Midler provides persuasive authority on this point. It is likely that the misappropriation of personality tort, at common law, extends to protect the unauthorized use of an artist’s voice. Assume it does. The next step is determining if the use was for commercial gain.

Commercial gain is required

The scope of the misappropriation of personality tort remains unclear. At the very least, it is constrained to “situations in which the name or image” (and presumably, the voice) of a celebrity or public figure is “used in the advertising or promotion of a defendant’s product or service, so as to imply that the celebrity endorses the activity of the defendant.”⁴⁷ In Gould Estate v. Stoddard Publishing Co., the Ontario Superior Court held that “the tort of appropriation of personality, is restricted to endorsement-type situations.”⁴⁸ Central to the “endorsement-type” situation, appears to be the requirement of commercial gain.⁴⁹

Typically, the tort protects against the misappropriation of one’s likeness in the context of advertisements and implied endorsements. If the voice-cloned music was used to advertise a product like Bette Midler’s was in the Ford commercial,⁵⁰ then the tort may be established. But it is unlikely that cloning a famous artist’s voice, alone, without commercial benefit or gain would satisfy the tort.

There is a strong argument, that when a song mimics an artist’s distinct voice, listeners naturally assume the artist has endorsed the music.⁵¹ Recall the “fake Drake” song, where TikTok user Ghostwriter977 “impeccably mimicked” the voices of Drake and The Weeknd.⁵² Indeed, the song went viral on TikTok and was uploaded onto Spotify, though it was removed shortly after.⁵³ It is possible that the “commercial” gain element would be satisfied, given that the song was available on Spotify, and may have generated royalties for Ghostwriter977. While an association, alone, may not be enough to give rise to an endorsement (as the law currently stands), if there is a commercial benefit to the defendant, the tort may be established.

But what if Ghostwriter977 did not post the “fake Drake” song on Spotify, and instead, posted it on X (formerly Twitter), or another non-monetized platform, resulting in no commercial gain at all? In Canada, the cause of action may be rendered toothless. In Horton v. Tim Donut Ltd., the Ontario Court of Appeal determined that using the defendant’s name and likeness to commission a portrait was neither exploitative nor commercial, despite the portrait being sold for a profit.⁵⁴ Part of the reason was that the proceeds from the poster sales were directed towards the Children’s Foundation – a charitable cause.⁵⁵ Since the use did not result in a personal commercial gain to the defendant, the court found that the common law tort of misappropriation of personality was not satisfied. While a charitable purpose alone will not invalidate the tort,⁵⁶ Horton demonstrates that in the absence of a direct commercial benefit to the defendant, the tort may not succeed.

Importantly, the tort also does not protect the use of a public figure’s name, image or likeness, when the public figure is the “subject” of the appropriation.⁵⁷ For example, where the appropriation is used to assist the public in learning more about a celebrity, like a biography, the appropriation will not constitute an improper commercial use of the celebrity’s personality. But it is unlikely that voice-cloned music places the artist as the subject of the appropriation, akin to a biography. Indeed, Ghostwriter977 posted the “fake Drake” song on social media, clarifying that he created it (implying that Drake did not).⁵⁸ Accordingly, Ghostwriter977 may argue that the song was a tribute to Drake, or otherwise placed Drake as the “subject” of the appropriation, because he did not imply that Drake created the music. But that argument should fail. While Ghostwriter977 took credit for creating the song, once it was publicized, control over it was beyond his reach. Certainly, when listeners hear a song in a well-known artist’s voice, they are likely to assume that the artist actually performed it.⁵⁹ This stands true regardless of whether Ghostwriter977 took credit for the creation when it was initially posted.

Ultimately, the existing misappropriation of personality tort may provide recourse, as it stands, for the commercial exploitation of an artist’s voice (though it has yet to be interpreted in that context.) Even so, the tort still falls short in protecting artists when there is no commercial benefit to the defendant. The limited Canadian case law on misappropriation of personality makes its application difficult to predict. While the tort shows promise, it needs significant improvements to effectively protect against voice-cloned music. As it stands, it is primarily applicable in commercial contexts; its application to voice cloning remains uncertain.

III. Revising the Misappropriation of Personality Tort

The Canadian privacy tort law regime has its shortcomings. Intrusion upon seclusion and public disclosure of private facts have limited prospects of success in the context of voice cloning. Those torts are better left alone. The tort of false light holds some promise and appears to provide a viable cause of action for voice-cloned artists, but the success of that claim remains entirely uncertain. The uncertainty of the claim is due to the lack of precedent. But the lack of precedent is because no one will bring a claim. The novelty of the tort both gives rise to the problem and is a consequence of it.

The misappropriation of personality tort provides a workable framework but fails to adequately protect against voice cloning as it stands. But amending the tort in an incremental way is the most viable option. For one, the tort is well-accepted in American jurisprudence, and across Canada. Provincial statutes explicitly codify the misappropriation of personality as a cause of action under invasion of privacy. As a result, the harm associated with the unauthorized use of an artist’s or celebrity’s name, image, or likeness is widely acknowledged. Amending the misappropriation of personality tort is favourable because it would provide a harmonized approach across all Canadian provinces. Indeed, the cause of action is distinct at common law; the provincial privacy statutes do not oust bringing a common law claim.⁶⁰ Either, or both, can be brought.⁶¹

Amending the privacy statutes would be no easy feat. And “few cases have fallen under their scope” anyways.⁶² Adjusting the scope of the misappropriation of personality tort at common law is far more practical. Judges can and will make incremental changes to the common law; to be realistic, they will do so prudently. The current misappropriation of personality tort has two main shortcomings that hinder its effectiveness for voice-cloned artists: the tort does not clearly protect one’s voice (or a cloned version of that voice), and it relies on a traditional concept of “commercial gain” to be actionable.

Under the amended misappropriation of personality tort, the plaintiff must demonstrate that the defendant, without consent: (1) misappropriated the name, image, likeness, or voice of the plaintiff (or a digital recreation thereof); (2) for gain, commercial or otherwise. If an artist has been voice-cloned, the plaintiff need not demonstrate actual proof of damage. The conduct is actionable so long as it can be shown that the voice-cloned music falsely associates the artist with the music.

1. Protection over an actual or simulated voice

The first case recognizing the common law misappropriation of personality tort in Canada defined the tort as a “remedy for the appropriation for commercial purposes of another’s likeness, voice or personality.”⁶³ Accordingly, it can be assumed that the common law tort does indeed extend to one’s voice. However, this protection requires clarification to address the specific harms related to voice cloning. The current challenge is no longer limited to the unauthorized use of one’s actual voice but must extend to a digital replication of that voice. Ghostwriter977, for instance, should not be allowed to evade liability by arguing that the voice was not Drake’s, but rather “fake Drake’s.” The definition of “voice” should be expanded accordingly.

In Tennessee, the ELVIS Act was introduced to provide “protections for songwriters, performers, and music industry professionals’ voice from the misuse of artificial intelligence (AI).”⁶⁴ The ELVIS Act expands the American right of publicity to protect one’s voice against commercial exploitation.⁶⁵ Of particular interest is the specific definition of “voice”, which has been broadened to include “a sound in a medium that is readily identifiable and attributable to a particular individual, regardless of whether the sound contains the actual voice or a simulation of the voice of the individual.”⁶⁶ A similar definition should be adopted to broaden the protection in the common law misappropriation of personality tort, to ensure that protection extends not only to an artist’s voice but also to a digital representation of that voice.

2. For the defendant’s advantage, commercial or otherwise

The existing misappropriation of personality tort requires that the defendant exploited the likeness of “the plaintiff for their commercial advantage.”⁶⁷ Respectfully, the tort misses the mark by focusing on the defendant’s commercial gain as the central harm.

The misappropriation of personality tort is indeed proprietary in nature. But it is intended to protect the “proprietary right in the exclusive marketing of one’s image and likeness.”⁶⁸ For artists in particular, the “proprietary” right to exclusively market themselves is breached when new, voice-cloned music is released without the artist’s consent – point blank, period. The right of an artist to market themselves includes their right to control their personal brand, which is shaped by the way their voice is experienced by the public. This is true, regardless of whether the defendant gained commercially from the misappropriation.

Some American states have broadened the scope of the tort. Under California’s equivalent common law right of publicity, a plaintiff must prove that the defendant used the plaintiff’s identity without consent, and the appropriation was for the defendant’s advantage, commercially or otherwise.⁶⁹ And if a defendant that voice clones an artist is not captured under the commercial gain requirement, the “otherwise” keeps the tort alive.

While Canadian courts have adhered to a more strictly commercial or endorsement context, the “otherwise” changes everything. California courts have recognized that it is “wholly unrealistic to deny that a name, likeness, or other attribute of identity can have commercial value.”⁷⁰ This too, was echoed by the U.S. District Court for Minnesota – “a name is commercially valuable as an endorsement of a product…because the public recognizes it and attributes…feats of skill…to that personality.”⁷¹ The commercial value of the artist’s likeness alone is enough to establish the tort. When any benefit is derived from the use of that identity (which has commercial value on its own), the benefit accruing to the defendant need not be “commercial,” too.

There is little reason to not extend this reasoning to an artist’s voice – the voice is commercially valuable on its own; that stands true. To be clear, a benefit that is non-commercial, in the context of voice cloning, is almost always established. The defendant gained the moment they used the artist’s voice without consent. The defendant gained by capitalizing off of the social position of the artist, their distinct voice, and their reputation. The tort is met accordingly. Non-commercial benefits are nonetheless tangible. For example, increased social visibility through media channels can arise from the creation of voice-cloned music. Cultural credibility, or “clout” as a “producer” or music creator is derived from the same. All of these benefits are gained by leveraging the artist’s years of hard work to cultivate a culturally recognizable and distinct musical voice. These are all benefits that may not be strictly commercial in nature but nonetheless support the plaintiff’s recovery.

Today, the nature of social media has made it so that one voice-cloned song can reach millions of people within a short time frame. When an artist’s voice is used to create new music, and that music goes viral, it is undeniable that the virality is, at least in part, a result of the use of the artist’s voice. For example, the fake Drake song did not go viral because it was a hit song (although it was certainly catchy); it went viral because it sounded like Drake. And the virality of the song alone, even without any commercial profit still would have resulted in tangible benefits for Ghostwriter977, through increased social media engagement and visibility. Certainly, Ghostwriter977 has amassed 250k followers on TikTok, and over 7k subscribers on YouTube.

In some countries, TikTok’s “Creator Fund” allows select users to earn revenue from their content.⁷² But Canada is not one of those countries; the Creator Fund is only available in the U.S., U.K., France, Germany, Spain, and Italy.⁷³ This means a Canadian TikToker could create another fake Drake song and not receive a single cent from it. Without recognition of the non-commercial benefits derived from the song, the plaintiff artist still may have no recourse (as it stands). But this is improper and warrants change – the defendant benefits, if not commercially, then “otherwise,” as soon as the defendant decides to use an artist’s voice without their consent to create new music. The tort should be amended accordingly.

3. Voice-cloned music is actionable without proof of damage

California’s common law right of publicity also requires the plaintiff to prove actual damage.⁷⁴ The Canadian common law is no different, the plaintiff bears the burden of establishing that there was actual damage.⁷⁵ In contrast, privacy statutes provide that the statutory cause of action is “actionable without proof of damage.”⁷⁶ While requiring proof of damage makes sense in most common law claims, it creates an unnecessary hurdle for artists whose voices are cloned without consent to create new music.

When voice-cloned music is used for commercial purposes, such as in an advertisement, akin to the Ford commercial imitating Bette Midler,⁷⁷ it is easier to prove economic damage. However, in non-commercial cases, such as when music is posted on a non-monetized platform, the harm may seem less tangible but remains significant. The damage stems from the false association between the artist and the voice-cloned content. Once that association is established, the artist should not be required to establish any further proof of damage.

In Athans, a plaintiff was a professional water skier and used a distinctive photograph of himself water-skiing.⁷⁸ The defendant reproduced a similar image in an advertisement for a children’s summer camp without using the plaintiff’s name or likeness. The plaintiff argued that the image’s resemblance to his own implied an endorsement. Though rejected due to a lack of evidence, the plaintiff’s argument remains compelling. He attempted to argue that his reputation “was likely to be tarnished commercially” because the image falsely associated him with the defendant’s “amateurish programme”, which would “bring him into disrepute as a professional athlete.”⁷⁹

Importantly, the court in Athans did not reject the premise of the plaintiff’s argument (a false association may indeed result in damage) but rather found that the plaintiff failed to establish the association at all, and accordingly, his reputation was unlikely to be tarnished. Even so, the court found that the defendant nonetheless liable for misappropriation of personality – not due to the false association, but because the defendant still benefited commercially from using the plaintiff’s likeness in their advertisements.

In the context of voice cloning, the association between the artist and the voice-cloned music is almost always going to be present. Indeed, the cloning of the artist’s voice is central to the wrongdoing. Accordingly, where there is evidence of an association between the artist’s voice, and the voice-cloned music, damage should be presumed for similar reasons as those advanced by the plaintiff in Athans. The use of an artist’s voice in an unauthorized musical work, is likely to damage the artist’s reputational and commercial interests, simply by falsely associating the artist with the voice-cloned song. Accordingly, the unauthorized publication of an artist’s voice that would reasonably lead listeners to believe the artist personally performed, recorded, or otherwise participated in the creation of a musical work is presumptively damaging. Accordingly, it should only be up to the plaintiff to demonstrate that the false association does indeed exist. Once established, no further proof of damage is required.

If Drake were to sue under the amended misappropriation of personality tort, he would need to demonstrate that: (i) Ghostwriter977 misappropriated his voice or used a synthetic recreation of it; and (ii) Ghostwriter977’s gained from the misappropriation, commercially or otherwise. Both elements are met. While Ghostwriter977’s conduct may have risen to the level of “commercial” gain to satisfy the existing tort, even if the gain was not commercial, the tort would be met under the amended version. And if Drake can prove that the cloned voice in the song created a false association between himself and the voice-cloned music in the eyes of the public, no further proof of damage would be needed. Drake would likely succeed in establishing that the fake Drake song, created a false association with him. And so, the tort is met.

Conversely, if Drake failed to demonstrate that the song created a false association in the eyes of the public, proof of actual damage would be required. As demonstrated in Athans, the tort can still succeed even if the association is not established, as long as the plaintiff can prove that the misappropriation resulted in other damage. The Ontario Superior Court in Athans calculated damages by determining the amount the plaintiff “ought reasonably to have received in the market for permission to publish the drawings.”⁸⁰ For the “fake Drake” song, perhaps this calculation could be derived from the amount Drake would have charged if Ghostwriter977 were to obtain his permission to be featured on the song. While unconfirmed, rumours suggest that Drake has historically charged nearly one million dollars for a feature.⁸¹ In an ideal world, Drake would be able to recover something similar. While hefty, the metric makes sense – the more famous an artist is, the more a false song would impact their reputation and, by extension, their position in the marketplace.⁸²

While misappropriation of personality claims will often result in economic damage, American courts have recognized that “the appropriation of the identity of a celebrity may induce humiliation, embarrassment, and mental distress,” too.⁸³ Canadian courts should recognize the same; damages for the amended tort should not be limited to economic harm. Damages may be substantial then, but the risk acts as a major deterrent. The best way to mitigate the harm is to prevent it from occurring, after all.

IV. Addressing Freedom of Expression Concerns

Canadian courts have recognized that limitations on the misappropriation of personality tort are necessary to preserve “society’s interest in free expression, ideas, newsworthy events, or matters of public interest.”⁸⁴ But these “limitations” are unlikely to arise in the context of voice-cloned music to begin with, and thus, would present no compelling basis to reject, or defend against liability.

To start, society’s interest in free expression has always been balanced against privacy concerns. In Aubry v. Éditions Vice-Versa Inc., the Supreme Court of Canada held that the right to creative expression must be balanced against the privacy interests of the person impacted.⁸⁵ There, the defendant photographer took an image of the plaintiff, who was sitting in a public area, and published it in a magazine without her consent. Given that the plaintiff was identifiable in the image, the Court reasoned that “the respondent’s right to protection of her image” prevailed over the appellant’s “right to publish the photograph of the respondent without first obtaining her permission.”⁸⁶ Central to the holding is that “artistic expression…cannot justify the infringement of the right to privacy it entails.”⁸⁷ Notably, the case was dealt with in reference to the Civil Code of Quebec,⁸⁸ where it is an invasion of privacy to use one’s “name, image, likeness or voice for a purpose other than the legitimate information of the public.” Albeit not the same, the cause of action is comparable to the amended misappropriation tort, assuming it no longer focuses solely on commercial gain.

While Aubry was made with reference to the Quebec Québec Charter of Human Rights and Freedoms,⁸⁹ the principle is equivalent here. Falsehoods (and by extension, voice-cloned music) remain protected under section 2(b) of the Canadian Charter of Rights and Freedoms but remain more susceptible to limitation.⁹⁰ The limitation here is justified. The creation of voice-cloned music may be inevitable. Indeed, it may be said that an artist has relinquished some expectation of privacy in their voice like the plaintiff in Aubry had a reduced expectation of privacy by sitting outside on the steps. However, the Supreme Court of Canada has affirmed that the plaintiff’s right to protect her public image outweighs the right to publish the image as a form of creative expression without her consent.  And similarly, an artist’s right to protect their public image (and voice) outweighs the right to publish voice-cloned music as a form of creative expression without consent. Thus, the amended tort strikes the right balance.

Further, the limitations arising from the amended tort will not stifle innovation; individuals are still free to create AI-generated music. MusicLM, for example, is a text-to-music generator created by Google that enables creators to make new music, but it does not endeavour to replicate the voice or style of a particular artist.⁹¹ MusicLM turns a simple text prompt into a melody and is anticipated to expand to “lyrics generation, along with...high-level song structure like introduction, verse, and chorus.”⁹² One may use MusicLM and will not be liable under the amended tort. The concern lies not in technological progress itself but in applications that exploit artists’ distinctive traits without their consent.

Some artists may indeed provide their consent willingly. Canadian artist Grimes, for instance, has stated that fans are “free to use [her] voice without penalty.”⁹³ She has invited “creators to use AI-generated versions of her voice to make new music,”⁹⁴ through her voice cloning software, Elf.Tech.⁹⁵ In exchange, Grimes wants a 50/50 split of any royalties generated through the commercial use of her voice.⁹⁶ For non-commercial uses, Grimes has relinquished control. To her, the “ultimate goal has always been to push boundaries” and to “see what happens even if it’s a bad outcome for [artists].”⁹⁷ And she believes there is not much to lose. In her words, “we are closer to cyborgs than we think”, and humans are already “fully integrated with technology.”⁹⁸

While some artists, like Grimes, may consent to the use of their voice to create new musical works, the same cannot be said for all artists. Those who do not consent to such use, deserve recourse when their voice is repurposed. And that recourse should be available; the amended misappropriation of personality tort provides just that.

V. Concluding Thoughts

The law hopes to be stable, but it must not stand still. This research has argued that an expansion of the misappropriation of personality tort is the most viable and incremental way to address voice cloning harms, particularly in a non-commercial context. While a song created by voice cloning can be viewed as a form of creative expression, the dissemination of voice-cloned music dilutes the very essence of an artist’s career – their voice.

The amended misappropriation of personality tort should: (1) expressly protect one’s voice, or a “mimic” thereof, and (2) be actionable when the appropriation provides a benefit or advantage to the defendant, commercially or otherwise. Once it is established that the misappropriation gave rise to a false association between the artist and the voice-cloned music, the tort should be actionable without further proof of actual damage.

The amended tort remains aligned with and does not unduly restrict freedom of expression under s. 2(b) of the Charter.⁹⁹ While technological advancements have undeniably made voice cloning more accessible, technology itself should not bear the blame. AI can certainly enhance creative expression, and there is a place for AI-generated musical works. However, the unauthorized replication of an artist’s voice to create new music without consent constitutes harm that must be actionable. Expanding the tort in this context is a balanced and necessary response.

Indeed, necessity is the mother of invention. But the amended tort is not immune from criticism. It too, is imperfect; this research does not attempt to resolve all of the difficulty. For one, identifying the responsible party behind a voice-cloned song can be challenging. Artists may not know who to sue. Many would argue that liability should rest with the individual who created the voice-cloned song.¹⁰⁰ Others might contend that responsibility should extend to the developers responsible for voice cloning technology. Some suggest that social media companies or streaming platforms that host and facilitate the distribution of these songs bear some accountability.¹⁰¹ There is no clear answer, but an artist may understandably wish to recover from a large corporation to maximize recovery.

Reputational concerns also play a role. A presumably wealthy, well-known artist suing an unknown person, might appear heavy-handed, regardless of legal merit. Perhaps this is more of a reason for artists to bring a claim against social media, streaming, and technology companies as opposed to the common user of voice cloning technology.¹⁰²

While tort law may provide injunctive relief, it is no secret that litigation is expensive, and courts move slowly. Tort law, then, may not provide the expedited remedy presumably desired. At the very least, however, tort law carries a certain level of intimidation. The increased viability of an artist’s claim alone might compel an adversary to settle.

And while not perfect, the end may very well justify the means.

Bibliography

Canadian Legislation

The Privacy Act, CCSM, c. P125, s. 2(1)

Privacy Act, RSBC 1996, c. 373, s. 1

The Privacy Act, RSS 1978, c. P-24, s. 2

Privacy Act, RSNL 1990, c P-22, s. 3

Canadian Case Law

Athans v. Can. Adventure Camps Ltd., 1977 ONSC 1255

Aubry v. Éditions Vice-Versa Inc., [1998] 1 SCR 591.

Boucher v. Wal-Mart Canada Corp., 2014 ONCA 419

Caplan v. Atas, 2021 ONSC 670

Dawe v. Nova Collection Services (Nfld.) Ltd., [1998] N.J. No. 22

Doe 464533 v. N.D., 2016 ONSC 541

E.S. v Shillington, 2021 ABQB 739

Gould Estate v. Stoddart Publishing Co. 1996 ONSC 8209

Grant v. Torstar Corp., 2009 SCC 61

Hay v Platinum Equities Inc., 2012 ABQB 204

Horton v. Tim Donut Ltd. [1997] O.J. No. 4154

Jones v. Tsige, 2012 ONCA 32

Joseph v. Daniels, 1986 BCSC 1106

Krouse v. Chrysler Canada Ltd. et al., 1973 ONCA 574

Parasiuk v. Canadian Newspapers Co. [1988] 2 W.W.R. 737 (Man. Q.B.)

Racki v. Racki, 2021 NSSC 46

R v. Zundel, [1992] 2 SCR 731

Salé v. Barr, 2003 ABQB 431

S.B. v. D.H., 2022 SKKB 216

Tucci v. Peoples Trust Company, 2020 BCCA 246

Yenovkian v Gulian, 2019 ONSC 7279

U.S. Case Law

Comedy III Prods. v. Saderup, 21 P.3d 797 (Cal. 2001)

Estate of Presley v. Russen, 513 F. Supp. 1339 (DNJ 1981) at 1354

Midler v. Ford Motor Co., 849 F.2d 460, 462 (9^th Cir. 1988)

Motschenbacher v. R.J. Reynolds Tobacco Co., 498 F.2d 821 (9^th Cir. 1974)

Uhlaender v. Henricksen, 316 F. Supp. 1277 (D. Minn. 1970)

U.S. Legislation & Government Documents

Tennessee, Office of the Governor, “Gov. Lee Signs Elvis Act into Law” (21 March 2024)

Ensuring Likeness Voice and Image Security (ELVIS) Act, Tenn Code Ann § 47-25-1105 (2024)

Secondary Sources: Journal Articles

Adam Goldenberg, Leah Strand, and Ambreena Ladhani, “Deep fakes: Bringing the tort of false light out of the shadows”, 42 Adv J No 3, 40-44

Amy M. Conroy, “Protecting Your Personality Rights in Canada: A Matter of Property or Privacy?”, (2012) 1:1 online: UWO J Leg Stud 3

Andrea Agostinelli et al, “MusicLM: Generating Music From Text” (January 26, 2023), arXiv.

Bryn Wells-Edwards, “What’s in a Voice? The Legal Implications of Voice Cloning” (2022) 64:4 Ariz L Rev 1213 at 1215.

G.H.L. Fridman, “The Law of Torts in Canada”, 3^rd ed (Toronto: Thomson Reuters, 2010).

Harnoorvir Singh Josan, “AI and Deepfake Voice Cloning: Innovation, Copyright and Artists’ Rights” (2023), Centre for International Governance Innovation, online (PDF).

Kavyasri Nagumotu, “Deepfakes Are Taking Over Social Media: Can the Law Keep Up?” (2023) 62:2 Franklin Pierce C Intellectual Property J 102.

Rachel Reed, “AI Created a Song Mimicking the Work of Drake and The Weeknd. What Does That Mean for Copyright Law?” (May 2, 2023), Harvard Law Today, online.

William Prosser, “Privacy” (1960) 48 Cal. L. Rev. 383 at 389.

Secondary Sources: Online Resources

Ben Rogerson, “Heart On My Sleeve, the song at the centre of the ‘Fake Drake’ AI storm, has been submitted for two Grammy Awards, but there could be a problem” (2023) Music Radar.

FineShare, “Jay Z AI Voice Generator” (2025).

@ghostwriter977, “my song ‘Heart on My Sleeve – ghostwriter” (April 16, 2023). X (formerly Twitter).

@Grimezsz, “Feel free to use my voice without penalty. I have no label and no legal bindings.” (23 April 2023), X (formerly Twitter).

@Grimezsz “The ultimate goal has always been to push boundaries rather than have a nice song” (24 April 2023), X (formerly Twitter).

Grimes, “International Music Summit Ibiza 2023: Keynote Interview” (4 May 2023), online: YouTube

Isaac Semple, “The Five Most Expensive Verses in Rap History” (4 November 2023) Hiphop Hero.

Jazz Monroe, “Grimes Unveils Software to Mimic Her Voice, Offering 50-50 Royalties for Commercial Use” (2023), Pitchfork.

Joe Coscarelli “Ghostwriter Returns With an A.I. Travis Scott Song, and Industry Allies” (2023). New York Times.

Lalal.ai, “AI Voice Generator: Generate Drake’s Voice Online” (2025)

Parrot AI, “Taylor Swift AI Voice” (2025)

Rachel Reed, “AI created a song mimicking the work of Drake and The Weeknd. What does that mean for copyright law?” (2023) Harvard Law Today.

R/popheads, “AI-generated Fake Drake/Weeknd collaboration ‘Heart on My Sleeve’ goes viral” (16 April 2023), online (Reddit Thread).

Vanessa Romo, “Grimes Invites Fans to Make Songs with an AI-Generated Version of Her Voice” (24 April 2023), NPR.

Endnotes