By: Dhruva Krishna
I. Virtual Avatars: Lazarus to Luke Skywalker, Deepfakes to Dystopia
Virtual avatars are digital recreations of an individual or their performance. Entertainers now regularly appear in all forms of media as virtual avatars–in video games, as holograms, and as younger versions of themselves in film.
COVID-19 has increased the prominence of virtual avatars. In April 2020, Travis Scott broke records for his Fortnite performance, drawing 12.3 million viewers and reportedly earning $20 million for a nine-minute digital concert. The virtual concert platform, Wave, has produced virtual events for artists like The Weeknd, Lindsey Sterling, John Legend, and others. The platform raised $30 million in Series B funding from investors like Scooter Braun, a music industry mogul, and Kevin Lin, one of the co-founders of Twitch, and recently announced partnerships with Tencent Music Entertainment, China’s leading music company.
It took four days for Lazarus to rise from the dead – entertainers might be resurrected faster. Some examples of celebrity “resurrections” include the touring Buddy Holly and Roy Orbison holograms, the recreation of Peter Cushing as Grand Moff Tarkin in Rogue One, or the Rob Kardashian hologram Kanye West “gifted” to Kim Kardashian.
These developments reveal a technological revolution underway. For instance, advancements in artificial intelligence (“AI”) technology, machine learning, and data processing have upended the way music is typically produced. AI can now make original, highly realistic creations by “learning” from existing work. Jukebox, an easy-to-use neural net created by OpenAI, “generates music, including rudimentary singing” in a variety of different styles by drawing on a dataset of 1.2 million songs.
This revolution has democratized access to extremely powerful technologies. People can create their own virtual avatars with little to no programming knowledge or cost. In 2021, a Star Wars fan recreated the appearance of Mark Hamill as Luke Skywalker in the popular show “The Mandalorian” using deepfake technology. The deepfake was heralded as a significant improvement over Disney’s virtual avatar.
The confluence of powerful technology and increased access can be destructive, however. Scholars, celebrities, and private individuals have variously articulated how the use of virtual avatars can effect immense personal, moral, and economic harm. For instance, nonconsensual pornographic deepfakes proliferate online with little ability to stop their spread. Additionally, the nation was in a state of paranoia over the use of deepfakes and virtual avatars to disrupt the 2020 election, leading to new content restrictions online.
At the crux of these issues is a legal status quo that does not afford sufficient protections. There are issues with enforcement, takedown, and the proliferation of harmful virtual avatars. Advocacy groups have pushed for expanding existing legal protections or creating new ones. More recently, these protections have focused on crafting digital replica and likeness laws.
II. New York’s Right of Publicity Exemplifies Digital Replica and Likeness Legislation Flaws
There have been federal and state efforts to create new digital replica and likeness laws. Such laws include mandatory disclaimers for manipulated media, banning virtual avatars during election periods, and creating causes of action for nonconsensual digital pornography.
There are common weaknesses amongst these efforts. First, these laws employ confusing language, creating overinclusive and underinclusive protections. Second, these laws have unclear and ineffective exceptions or disclaimer requirements. Ultimately, these laws are likely to deter valuable innovation in new technological fields.
On November 30, 2020, Governor Andrew Cuomo signed an updated New York right of publicity statute into law. The statute ended the battle to recognize New York’s right of publicity postmortem protections, much like its California counterpart. Yet, New York’s right of publicity law exemplifies the weaknesses of digital replica and likeness laws.
A) Deceased Performer, Deceased Personality, and Digital Replica Protection Flaws
New York’s new right of publicity law creates new provisions encompassing virtual avatars, particularly protections covering “deceased performers,” “deceased personalities,” and “digital replicas.” Specifically, they provide new avenues of imposing civil liability against individuals using an artists’ works and likenesses. These protections last forty years after an individual’s death, as set out in §52f-7(c)(8). However, the protections provided by these provisions, and the way that they intersect, are confusing.
i) “Deceased Performer” and “Deceased Personalities”
Take, first, New York’s new provision, §50-f(1)(a), defining “deceased performer.” These are performers who “for gain or livelihood…regularly engaged in acting, singing, dancing, or playing a musical instrument.” This provision encompasses modern virtual avatars like Roy Orbison, Michael Jackson, Whitney Houston, and more.
However, this language is overinclusive by using a “for gain or livelihood” threshold, which sweeps in much more than just professional entertainers. If an attorney performs in a band, would they fall under the provision? If a teenager posted videos of themselves dancing on social media sites to sell products, would they fall under this definition? The additional qualification of “regularly engaged” does not provide guidance. Is this once a week? Twice a year? Three times a month? More private individuals may now have rights of publicity, which may not have been within the contemplation of the statute.
Additionally, entertainers who may not fall under the deceased performer definition, such as athletes or politicians, would be encompassed by the definition of “deceased personality” in §50-f(1)(b), which covers any person “whose name, voice, signature, photograph, or likeness has commercial value” at their time of death or because of their death. It does not matter if the individual actually utilized these aspects during their lifetime for gain.
The commercial gain provision is similarly overinclusive. The qualification of “commercial gain” is not defined. This vagueness is troublesome in the age of social media, when many people are using their likeness for monetization. Consider a local yoga teacher who posts weekly tutorials on her Instagram page, with just several hundred local followers. After her death, the yoga teacher’s estate demonstrates that her posts drew several new customers annually. This yoga teacher might have a deceased personality right.
The over inclusivity of both provisions raises thorny questions. The statute opens the door to a slew of private individuals who may be able to claim virtual avatar rights, potentially forcing courts to draw lines on what “commercial gain” means. Evidentiary questions may be raised on how to best demonstrate that an individual used their likeness for gain or livelihood. Even if these questions are decided conclusively, the statute proposes a registration system for postmortem rights of publicity in §50f-7, which requires filing a form with the secretary of state, making the relevant claim a public record, and creating a website with this information available. Importantly, if one does not register their right, they may not have a cause of action. If one is unsure if their likeness would be covered, this creates timing and resource issues regarding accurate registration. Additionally, this registration system may become overwhelmed with requests if the threshold for commercial gain is too broad. Thus, further guidance is needed to clarify this ambiguous language.
ii) Digital Replicas
New York provides a new provision, §50-f(1)(c), defining digital replicas. A digital replica is a “newly created original, computer-generated, electronic performance of an individual in a separate and newly created, original expressive sound recording or audiovisual work in which the individual did not actually perform.” However, the replica must be “so realistic that a reasonable observer would believe it is a performance by the individual being portrayed and no other individual.”
This definition favors virtual avatar creators. Given the state of technology, a combination “reasonable observer”/“so realistic” standard seemingly protects against the most harmful infringing uses. A hologram of four dinosaurs performing Beatles songs would not fall under the digital replica provision as no reasonable observer would think The Beatles are performing. In contrast, a highly realistic deepfake of The Beatles performing would be subject to liability.
Changing expectations of technology and artists may change this provision’s application. As deepfake and highly realistic avatars become more common, a “reasonable observer” may be less likely to believe any virtual avatar is an authentic or realistic performance. Additionally, a performance’s realism may depend on the entertainer. A highly realistic avatar of The Beatles performing death-metal may still seem unrealistic enough to escape liability. But what about Brian Wilson of The Beach Boys performing a rap song? Or Johnny Rotten of The Sex Pistols performing new-wave music with jazz-legend Tony Williams and virtuosic-guitarist Steve Vai?
Furthermore, defining what a digital replica does not encompass raises questions. The provision excludes sound recording remasters and the “making or duplication of another recording that consists entirely of the independent fixation of other sounds, even if such sounds imitate or simulate the voice of the individual” [emphasis added]. What constitutes “other sounds?” Would OpenAI’s Jukebox neural network qualify under this provision? This may become a complex evidentiary issue, raising questions such as whether a network’s database incorporated an individual’s voice or used others to form a composite.
iii) Liability Under the New Statute is Expansive
Putting these various definitions together, the statute generally creates two avenues of liability under §50-f(2). The statute limits digital replica liability while providing expansive liability for deceased personalities.
Under §50-f(2)(b), the statute outlines limited liability for improper use of digital replicas. Any person, “who uses a deceased performer’s digital replica in a scripted audiovisual work as a fictional character or for the live performance of a musical work shall be liable…if the use is likely to deceive the public into thinking it was authorized by the person or persons [who can authorize such uses and did not provide their consent].” By limiting liability for digital replicas to deceased performers, the rights of individuals who did not “regularly engage” in “acting, singing, dancing, or playing a musical instrument” for “gain or livelihood” would not be protected.
The statute favors virtual avatar creators by requiring that the public think the use was authorized. Thus, two separate thresholds must be met: first, under the digital replica definition, the replica must be highly realistic to be believable to a reasonable observer; second, even if this digital replica threshold is met, the public must think the use was authorized. Generally, these standards would seem to align. A highly realistic Beatles virtual avatar would likely cause the public to believe the use was authorized.
The protections for deceased personality are much more expansive under §50f-2(a). Any “person who uses a deceased personality’s name voice, signature, photograph, or likeness, in any manner, on or in products, merchandise, or goods, or for purposes of advertising or selling, or soliciting purchases of, products, merchandise, goods, or service, without…[the person(s)’s consent] shall be liable…” [emphasis added]. Deceased personalities encompass a much wider category of individuals than deceased performers. Also, there is no realistic or reasonable limiting standard. Under this provision, a poor quality or sufficiently different avatar of the Beatles may be encompassed, even if it does not fall under the digital replica provision.
B) Confusing Exceptions
The New York bill contains many confusing exceptions. For instance, it limits liability for deceased personality avatars by excluding, in §50f-2d(i), literary works, works of art, parody, satire, and entertainment. Additionally, §50f-2(d)(iii) grants a newsworthiness exception and §50f-2(d)(iv) grants an exception for protected content that has sponsorship or commercial funding. However, it “shall be a question of fact whether or not the use of the deceased personality’s…[likeness] was so directly connected with the commercial sponsorship or with the paid advertising or product placement as to constitute a use for which consent is required…” §50f-2(e) makes a similar point, stating that no exception will be found if “the claimant proves that this issue is so directly connected with a product, article of merchandise, good, or service as to constitute an act of advertising, selling, or soliciting purchases…without prior consent for the use…” As experts have noted, branded content, sponsored posts, and blurry lines between authentic posts and paid advertising may create confusions for courts.
The bill also creates confusing exceptions for deceased performers and digital replicas. §50f-2(b) states that a digital replica is not encompassed if “the person making such use provides a conspicuous disclaimer in the credits of the scripted audiovisual work, and in any related advertisement in which the digital replica appears,” stating that the digital replica was not “authorized.” However, what constitutes “conspicuous?” Artists may try and construe this narrowly, whereas avatar creators may try and construe this broadly. If the disclaimer was boilerplate legalese at the end of a long show and delivered as the audience was leaving the amphitheater, would this count as conspicuous?
The bill creates carveouts within the exceptions for digital replicas. §50f-2(d)(ii) also creates an exception for newsworthiness for digital replicas. There is no liability “if the work is of parody, satire, commentary, or criticism; works of political or newsworthy value, or similar works…regardless of the degree of fictionalization; a representation of a deceased performer as himself or herself, regardless of the degree of fictionalization, except in a live performance of a musical work…” [emphasis added].
This carveout is confusing. Suppose a creator makes a Charlie Daniels virtual avatar that meets the bill’s requirements (preset script, highly realistic, etc.). This avatar goes on an a profanity-laced tirade about Democrats and Republicans before screaming “Never Vote Socialist.” Put aside if this is a satire, parody, commentary, criticism, or other generally protected work. For artists, the statute will still be overinclusive—if there is no “live performance of a musical work,” any representation of the deceased performer as “himself or herself” may be allowed. This seems to circumvent the purpose of the protection itself.
However, the provision also creates confusing standards for virtual avatar creators. Suppose that the creators underscore the avatar’s performance with “America The Beautiful”. Or, at the end of the tirade, the avatar plays a five-second-fiddle snippet of the “Star-Spangled Banner”. Are these sufficient elements to constitute a “live performance of a musical work”?
Digital replicas and digital performer rights are also subject to the exceptions of §50f-2(d)(iv) and §50f-2(e). As noted, these create confusing standards in the social media era.
C) Confusing Standards Can Harm Valuable Technology
The New York law is rife with ambiguities and confusing standards – sloppy lawmaking that can deter the development of valuable technology. As claims and lawsuits are brought under this law, new standards that widely enforce or constrain these provisions can have huge effects. If a court ruled that “commercial gain” constituted any monetization of one’s acting or musical performance, this would seemingly open the floodgates for possible claims. However, if “commercial gain” is a high threshold, then it could block many viable claims.
Additionally, judicial decisions could chill the development of valuable AI, machine learning, and virtual avatar technology. If wide-ranging regulations are brought into this space, creators may be unwilling to innovate or dedicate time to new creations for fear of litigation. The technology itself may move away from the entertainment space, hampering beneficial uses.
D) Postscript on Other Aspects of the New York Bill
Other aspects of the bill are ancillary to its application in the entertainment space—but could still prove very important. First, the New York bill creates a separate provision, §52c, to handle nonconsensual, sexually explicit depictions. The issue of nonconsensual sexual virtual avatars is serious, daunting, and presents life-changing harm to individuals involved. I commend the New York legislature for taking a stance against these uses. For those interested in these issues, I highly recommend the work of Danielle Citron.
Second, the New York bill explicitly recognizes the right of publicity as a transferable, descendible, property right. This is a unique categorization and is debated within the right of publicity sphere. Others have categorized the right of publicity as a privacy right, while others have categorized it as an intellectual property right. As virtual avatar legislation becomes increasingly commonplace, the different ways these rights may be categorized could create disconnects and conflicts among different legal frameworks.
III. Potential Areas of Reform and Research
Given the importance of legislation and the tension in this space, there are many areas of possible reform. First, stakeholders should explore enforcement mechanisms outside judicial decision making. One possible area is analyzing online platform practices. Platforms spread content, especially virtual avatars, and play a powerful role in curbing abusive uses. To that end, I have advocated that transparency disclosures and collaboration, rather than the removal of important protections like Section 230, would be the best way to spur improvements. Additionally, there seems to be a research gap in how different platforms create, define, and apply their own community standards to harms created by virtual avatars.
Second, stakeholders should recognize how existing law may be adjusted, reformed, limited, and encompassed by new technologies. Many individuals may have copyright, trademark, right of publicity, false light, and defamation claims against virtual avatars. However, these claims may also be limited and—especially for sexually explicit harms—may not provide remedies to prevent the spread of destructive avatars.
Third, entertainers and advocates should recognize the shift that is occurring within technological spaces. Artificial intelligence and its increased accessibility have already, and will continue, to change the world. Continuing to take a nuanced approach to this space ensures that new technology and rightsholders are respected. Stakeholders should consider if a substantive new body of law, paradigms, and rights surrounding virtual avatars will be created.
Dhruva Krishna is a recent graduate of UCLA School of Law. He will be entering corporate practice and has a passion for entertainment and technology.