House’s Draft AI Bill Risks Loss of Control over Our Own Voices and LikenessesBy Jennifer E. Rothman
January 23, 2024
***UPDATED TO REFLECT INTRODUCTION TO CONGRESS
On January 9, 2024, Representative Maria Salazar circulated a draft of a bill to address voice and likeness rights in the context of AI. The bill is titled “No Artificial Intelligence Fake Replicas and Unauthorized Duplications Act of 2024,” shorthanded as the “No AI FRAUD Act.” Unfortunately, despite its good intentions, the draft bill would make fake replicas more, rather than less, likely. Since I first posted this, the bill was introduced into Congress as H.R. 6943 with a bipartisan set of seven co-sponsors.
The bill like the Senate’s discussion draft of its No FAKES Act and associated one-pager highlights the threats that “recent advancements in artificial intelligence (AI) technology” pose to our ability to protect against unauthorized uses of our voices and likenesses. The findings section of the draft highlights the AI-generated “Heart on My Sleeve” song which initially generated a buzz as a new joint venture by hit recording artists Drake and The Weeknd, when it was not. The draft notes other examples of recording artists’ voices being used in unauthorized new songs. The Findings section of the draft also points to the dental plan ad that used an unauthorized image and seeming performance by actor Tom Hanks. It additionally notes the use of this technology to create “false, nonconsensual intimate images” of high school girls in New Jersey and elsewhere. As I have noted elsewhere each of these acts, likely violates already existing state right of publicity laws and in the context of the “Heart on My Sleeve” and the Tom Hanks ad also violate federal and state trademark and unfair competition laws.
This proposed House bill would create a federal “property” right to a person’s likeness and voice, regardless of whether the person is dead or alive. A person’s voice is defined broadly as including a person’s “actual voice” or a “simulation” that is “readily identifiable from the sound of the voice or simulation of the voice, or from other information displayed in connection therewith.” Similarly, a person’s “likeness” includes an “actual or simulated image or likeness of an individual . . . that is readily identifiable as the individual by virtue of the individual’s face, likeness, or other distinguishing characteristic, or from other information displayed in connection with the likeness.”
Violation. The law would be violated when “any person or entity who, in any manner affecting interstate or foreign commerce . . ., and without consent of the individual holding the voice or likeness rights affected thereby” does one of the following things:
- “distributes, transmits, or makes available to the public a personalized cloning service”—such a service is defined as “an algorithm, software, tool, or other technology, service, or device the primary purpose or function of which is to produce one or more digital voice replicas or digital depictions of particular, identified individuals.”
- “publishes, performs, transmits, or otherwise makes available to the public a digital voice replica or digital depiction with knowledge that . . .it was not authorized” by the party who holds those rights
- “materially contributes to, directs, or otherwise facilitates” any of the above conduct
The bill provides for a $50,000 statutory damage for cloning services or recovery of actual damages and attributable profits. The bill provides for a $5,000 statutory damage for digital depictions and digital voice replicas or recovery of actual damages and attributable profits. It also provides for punitive damages and reasonable attorneys’ fees. It seems like attorneys’ fees are only awardable to “injured” parties so prevailing defendants appear not able to recover.
The bill restricts liability for “negligible” harm—though it is not clear what would qualify, if anything, in that category. The bill emphasizes that harm includes “financial or physical injury” or simply “elevated risk” of either, “severe emotional distress,” and “deception of the public, a court, or tribunal.” Given that third-parties under the terms of the bill can own another’s voice and likeness and do whatever they want with them, it seems like the express harm to be remedied by the bill is substantially caused by the bill itself which allows third-party owners of a person’s voice or likeness to deceive the public into thinking the person actually performed or specifically authorized the digital performance.
The bill also suggests that “any digital depiction or digital voice replica which includes child sexual abuse material, is sexually explicit, or includes intimate images” is per se harmful. I note that the term “intimate images” is not defined in the bill.
Confusingly, the bill in C(2)(E) appears to allow both the underlying identity-holder and an owner (or licensee) of their rights to bring suit. This sets up a potential conflict where—if the transferability provision discussed below is allowed—an identity-holder and the owner of their voice and likeness could sue a party authorized to use the person’s voice and likeness by the other or to sue one another. This raises the identity-thicket problem, that I have identified elsewhere to a new level. See Jennifer E. Rothman, Navigating the Identity Thicket, 135 Harvard L. Rev. 1271 (2022).
Like the No FAKES Act—the party whose interests appear to be primarily served by this proposed legislation is the record labels and recording industry, not the performing artists highlighted in the findings. Subpart C(2)(E) demonstrates this with a special provision that provides that anyone who has an “exclusive personal services” contract with a “recording artist or an exclusive license to distribute sound recordings that capture the individual’s audio performances” can also bring suit.
Transferability. The bill dangerously makes this new federal property right in a person’s likeness and voice “freely transferable,” meaning that a person or entity other than the identity-holder themselves can own that person’s voice and likeness forever and in any context. A chilling and likely unconstitutional prospect that I have repeatedly warned against. See, e.g., Jennifer E. Rothman, The Right of Publicity: Privacy Reimagined for a Public World (Harvard Univ. Press 2018); Jennifer E. Rothman, The Inalienable Right of Publicity, 101 Georgetown L.J. 185 (2012).
The bill limits authorized “digital depiction[s]” or “digital voice replica[s]” to those where a person was represented by counsel or is over the age of 18 unless a court approves the agreement or if a collective bargaining agreement applies.
These are woefully inadequate protections. First, no one should be able to own another person’s likeness or voice regardless of representation or court review. Second, this allows unions to authorize third-party ownership of all of their members’ voice and likenesses. Third, given the expense of counsel and the lack of bargaining power and sophistication of many parties, these provisions will do little to stop such transfers.
Finally, and of great significance to the purported goals sets forth at the outset of the discussion draft, allowing such transfers will make it more likely that unauthorized uses of a person’s voice and likeness will be circulated and that the public will no longer be able to tell when a person actually performed or agreed to any specific use of their voice or likeness. For example, a record label could get in perpetuity rights to a person’s voice and likeness and simply replicate that person’s voice over and over again, creating new songs that the person never sang or specifically authorized. Record labels might benefit from this, but society as a whole will be much the poorer.
Postmortem Rights. The bill would extend a postmortem period of rights to “executors, heirs, transferees, or devisees” for every single dead person (“regardless of whether the individual has died before [the bill’s] effective date”). The bill provides an initial postmortem period of ten years after death but then is unclear on whether it continues forever thereafter if it doesn’t “terminate.” The postmortem rights can then seemingly terminate if after twelve years (initial ten plus two) there is a showing of no use for “commercial purposes” of the person’s identity by any of the designated individual’s above or if they are all dead.
This is a very odd provision. The focus on commercial use, appears to value and incentivize profiting from and commercializing the dead rather than protecting the grieving process of the survivors or the dignity of the deceased. In fact, given the way estate taxes work this exacerbates the problem of forcing the commercialization of the dead even when the deceased and their families did not want to commercialize them. See, e.g., Jennifer E. Rothman, “Mixed Victory for Jackson Estate in Tax Court” (discussing this issue).
Defenses and Limitations. Like the circulated No FAKES Act in the Senate, this house bill makes clear that a disclaimer revealing that the depiction, voice replica, or cloning service is not authorized is not a defense.
Interestingly, the bill suggests that the First Amendment is a defense to this proposed cause of action—which of course it is—but then tries to hem in such a defense by defining how such a defense should be understood in this context. Specifically, the draft proposes a likely unconstitutionally vague and underprotective balancing test in which the “public interest in access to the use shall be balanced against the intellectual property interest in the voice or likeness.” The bill then sets forth a number of factors to consider, including whether the use is “commercial”— a term which it does not define—whether the use is “necessary for and relevant to the primary expressive purpose of the work in which the use appears,” and whether the “use competes with or otherwise adversely affects the value of the work of the owner or licensee of the voice or likeness rights at issue.”
These are extreme limits on free speech. The First Amendment does not require a showing that a use of a person’s voice or likeness must be “necessary.” And determining what is meant by necessary is a challenging business. Further, how are courts going to determine the “primary expressive purpose” of using a person’s identity in historical novels, docudramas, biographies, or works of utter fiction that refer to real people in connection with this inquiry of necessity?
Another section on limitations suggests that another relevant consideration is whether the use is “transformative.” This term is not defined and although it has been adopted by some courts in right of publicity cases in the context of First Amendment defenses, it originated in the context of copyright’s fair use defense, and likely has been narrowed in the publicity context as well, after the Supreme Court’s decision in The Andy Warhol Foundation for the Visual Arts v. Goldsmith, 598 U.S. 508 (2023). The bill also suggests that uses are allowed if “constitutionally protected commentary on a matter of public concern.” This version of a limit essentially leads us back to the First Amendment analysis that is quite muddled in the context of right of publicity cases across the federal circuit courts and states. See Robert C. Post & Jennifer E. Rothman, The First Amendment and the Right(s) of Publicity, 130 Yale L. J. 86 (2020).
No Preemption. Further exacerbating the identity thicket problem with this legislation, the bill does not preempt state (or federal) laws that also operate in this space suggesting that many parties may be able to claim rights over the same person’s voice and likeness. Leaving in place longstanding precedents that are working is a good thing, but adding to the muddle of conflicting rights over a person’s identity is not.
Section 230. The bill explicitly qualifies the rights provided as “intellectual property” for purposes of Section 230 meaning that internet service providers will not be immunized from liability for third-party speech on their platforms that violates the bill. The bill still leaves open the question of how to treat state right of publicity laws about which there is a split.
* * *
In sum, this discussion draft is in the early stages and no doubt will go through many iterations and improvements should it proceed. It is essential that in the process, Congress take more seriously (1) the threat posed by transferring a person’s rights to their own voice or likeness to another person or entity; and (2) more clearly identify the objectives of this legislation and limit its scope so as not to make things worse for performers, ordinary people, creators, and the protection of truth and authenticity.