Copyright Office Calls for Congressional Action on Digital Replicas
By Jennifer E. RothmanSeptember 12, 2024
Late this summer, the U.S. Copyright Office released the first of its planned reports on Copyright and Artificial Intelligence. This one, designated as Part 1, addresses “Digital Replicas”. It was notably released the same day that members of the Senate introduced a bipartisan digital replica bill. The Report generally supports a federal digital replica law and substantively tracks aspects of the proposed NO FAKES Act while parting ways on some issues.
It might be surprising to some that the Copyright Office’s first report on Copyright and AI isn’t really about copyright. The Report and proposed legislation focus on protecting a person’s voice and likeness in the context of AI-generated digital replicas. People’s voices and likenesses are not copyrightable, even though they can be captured in copyrightable works. But the speed with which new laws and bills are being introduced to address unauthorized uses of people’s voices and likenesses likely drove the Office to prioritize this report. This is not the first time the Office has considered the right of publicity and related laws that protect against unauthorized uses of a person’s identity. In its 2019 report examining moral rights for authors in the U.S., titled Authors, Attribution, and Integrity: Examining Moral Rights in the United States, the Office called for a federal right of publicity law to address the lack of uniformity among state right of publicity laws.
I commend the Office and its staff for the substantial work they put into preparing this new Report on Digital Replicas. The Office received more than 10,000 comments in response to their August 2023 Notice of Inquiry, and the end-result shows the seriousness with which the Office considered them. The Report is thoughtful, detailed, and helpful. There is much to agree with in the Report’s analysis and recommendations, but also some things that it gets wrong, as well as some recommendations with which I disagree.
The report runs 57 pages, so the following provides some highlights.
A useful thing to keep in mind is that the Office notes at the outset of the Report that its focus is on “the Constitutional goal of promoting creativity in order to ultimately benefit the public.” As the Registrar of Copyrights and the Director of the U.S. Copyright Office highlights in her Foreword to the report, “[t]his requires an appropriate balance, enabling technology to move forward while ensuring that human creativity continues to thrive.” This framing makes absolute sense for the Copyright Office but is not the only important frame when it comes to considering digital replica rights or broader rights over a person’s identity, including their voice and likeness. The protection of these interests goes far beyond the incentive rationale of copyright law and the goal of promoting creativity. Focusing solely on generating content and enhancing creativity will underprotect the autonomy and dignity of individuals. Personality or publicity rights are not external to the underlying people in the same way a book or song is in relation to its author. Accordingly, we need to think about intellectual property rights in people differently than we think about intellectual property rights in expressive works.
I will primarily focus on the Report’s recommendations for a federal digital replica law, and only where relevant note some disagreements with regard to the supporting background section of the report.
Targeted Digital Replica Law
The Office recommends adoption of a federal digital replica law rather than a “broader ‘name, image, and likeness” law. This is in part so as not to unsettle state law, but also likely in part to track the focus of the Senate bill that was introduced the same day as the Copyright Office released its Report. Notably, the Copyright Office had previously advocated for a broader federal right of publicity law to address the conflicting set of state laws that cover a person’s name, voice, and likeness. However, recognizing the danger of unsettling longstanding state law, the Office now suggests a narrower provision.
This narrower approach is simpler and perhaps more likely to pass. But even a narrower federal law that does not preempt state laws will add yet another layer of potentially conflicting rights over a person’s voice and likeness to the current “identity thicket.” Many state and federal laws already extend protection against unauthorized uses of a person’s voice or likeness, including when used in computer-generated digital replicas. It is not clear how conflicts between these overlapping rights will be resolved in the context of applying a federal digital replica right. Given the many possible claimants allowed in the introduced Senate NO FAKES Act, this problem could be quite significant.
The Office deems passage of “new federal legislation” to address digital replicas as “urgently needed.” This claim of urgency undersells the role of existing law in addressing these claims—state publicity and privacy laws, negligence, intentional infliction of emotional distress claims, and federal and state false endorsement, unfair competition, and trademark laws all likely apply to digital replica claims. There are numerous lawsuits pending under existing laws related to uses of generative AI and it is too early to know how they will come out. The technology is also evolving, so any laws must be designed with these changes in mind. Rushing through a federal law that has significant negative consequences would be worse than letting these lawsuits and the development of technological guardrails play out further. To be clear I do not oppose a federal law in principle, but it cannot be considered in a vacuum. The details matter.
Everyone Should Have a Digital Replica Right
The Office importantly and appropriately advocates that everyone, not just celebrities, should be able to bring claims and limit unauthorized uses of their voices and likenesses in the context of digital replicas. To facilitate this, the Office supports the inclusion of statutory damages and fee-shifting. While most states allow claims by ordinary people under their right of publicity laws, some states have limited claims to those whose identities have commercial value and many do not provide statutory damages or fee-shifting.
Postmortem Rights Not Needed and the Living Should be Prioritized
The Office astutely notes that the focus of a digital replica right should be to “prioritize the protection of the livelihoods of working artists, the dignity of living persons, and the security of the public from fraud and misinformation.” The Report concludes that a “postmortem term is not necessary” to achieve these goals. This parts way with the bills currently being floated in Congress. Perhaps because of this, the Report observes that one could make a “reasonable argument for allowing heirs to control the use of and benefit from a deceased individual’s persona that had commercial value at the time of death.”
The Report suggests that if a postmortem digital replica right is recognized, it should be “limited in duration” (to twenty years or less) and perhaps allowed to be extended if the right is commercialized. Again, this interestingly tracks the Senate bill released the same day as the Report. The Report does not explain why such a postmortem right is justified nor why it should focus on commercializing the dead. Such a focus supports a troubling effort to build a market in dead people that doesn’t actually further the wishes of the dead or their survivors or serve the public good. Instead, postmortem provisions should focus on the preferences of the deceased and the well-being of their relatives, and should not vest in unrelated third-parties. For a more detailed discussion of these issues see Anita L. Allen & Jennifer E. Rothman, Postmortem Privacy (forthcoming in the Michigan Law Review (2024)). (I note that the Report suggests that privacy rights “do not survive death.” This is a common claim but turns out to be incorrect and therefore does not answer the question of whether (or how) we should extend postmortem rights in a person’s identity after death. For more on this issue see id.
Digital Replica Right Applies to Noncommercial Uses
The Office wisely advocates that the digital replica right should extend to noncommercial uses, as many harms flow from uses that have nothing to do with making money or uses on or in merchandise or advertising.
Limits on Liability
The Office does not think there should be liability for merely creating a digital replica. The Office suggests that liability should require the “distribution” or “making available” to the public of an unauthorized digital replica. Creation could form the basis of liability if it is part of a broader distribution scheme. The Report emphasizes the value of allowing the creation of digital replicas for personal use as part of “personal entertainment” or “an artist’s experimental process.”
The Office recommends requiring an actual knowledge standard for liability, as the NO FAKES Act does. The Report rejects arguments that liability should turn on an intent-to-deceive standard, noting that this is too “high a barrier” given the harm that stems from the circulation of “unauthorized replicas.”
Limits on Licensing
The Office recognizes that the digital replica right should not be assignable during a person’s lifetime and suggests that “guardrails” are required for the licensing of such rights. The Office recommends that licenses be for a limited duration—it suggests 5-10 years, which as I discuss elsewhere is far too long, particularly without significant limits on scope and ongoing supervision by an identity-holder.
To address some of these concerns, the Office recommends that a digital replica license be required to “ensure” that individuals have “adequate knowledge and full disclosure of intended uses.” As I wrote in my analysis of the NO FAKES Act, this knowledge must be specific and therefore any broad licenses must require ongoing control and supervision by the individual. Given the Office’s important observation that the digital replica right is in part directed at protecting “the public from fraud and misinformation” this ongoing control is also essential, as is the disclosure that a performance is by a replica rather than authentic. Such disclosure is not something, however, that the Report considers. The Office also notes the importance of providing additional protection for minors.
The Office highlights that works that are created during the period of licensing should continue to be capable of use and distribution even after the expiration of licensing. For example, a sound recording that uses a digital replica should continue to be capable of distribution after a license expires, but new uses of the digital replica would not be permissible after the license’s expiration. I note that this determination would depend at least in part on the terms of a licensing agreement unless a law codifies specific parameters. The recommendation does not address the question of licensed samples in new works.
Allow Secondary Liability and Exclude Replica Right from Section 230
The Office advocates allowing secondary liability for violation of a digital replica law using traditional tort analysis for contributory and vicarious liability. In the context of online service providers, the Office recommends that OSPs should not be immunized by CDA Section 230 from secondary liability. The Office concludes that this is important to “encourage prompt removal of unauthorized digital replicas.” I agree. The potential OSP liability is mitigated by the Office’s call for a notice and takedown process that provides a safe harbor until an OSP is put on actual notice. This approach is included in the Senate NO FAKES Act.
Calls for a Fair Use Exception to Liability
The Report agrees with the expressed “First Amendment [c]oncerns” about adopting a federal digital replica right. The Report also accurately reports the uncertainty and confusion surrounding the application of the First Amendment to right of publicity claims. (The Report quotes Bill McGeveran’s take that this area of the law is akin to a “dumpster fire.” See William McGeveran, Selfmarks, 56 Hous. L. Rev. 333, 362 (2018); see also Robert C. Post & Jennifer E. Rothman, The First Amendment and the Right(s) of Publicity, 130 Yale L.J. 86 (2020).
The Office suggests that an exemption-based approach, for example, exempting “expressive works,” would be problematically under- and overinclusive. The Office accurately notes that deepfake pornography and musical clones could be protected under some proposed exemptions. The Report also observes that the “traditional [exemption] categories such as ‘news’ or ‘public affairs’” are breaking down “in today’s online environment.” Instead of adopting such exemptions, the Office recommends a copyright-like “balancing” fair-use-style approach that could take into consideration a number of factors, including the “purpose of the use, including whether it is commercial; its expressive or political nature; the relevance of the digital replica to the purpose of the use; whether the use is intentionally deceptive; whether the replica was labeled; the extent of harm caused; and the good faith of the user.” This is the best articulation I have seen of what sorts of factors might be relevant. The Office persuasively argues that this approach would still be highly speech-protective, even if somewhat less predictable (similar to copyright’s own fair use doctrine). Notably, the Office (in a footnote, number 284 for those looking) observes that using “unauthorized digital replicas of deceased individuals” should be held fair under these factors when used to “portray[]” real individuals in “unrealistic, fictional contexts.” The Office points to the use of a time machine to encounter historical figures in a “remake of the movie Bill & Ted’s Excellent Adventure.”
Interestingly, the Report recognizes that producing a work in the “style” of someone should not violate this digital replica right. At the same time, the Office suggests an amendment to the Copyright Act to clarify that “section 114(b) . . . does not preempt or conflict with laws restriction unauthorized voice digital replicas.” Section 114(b) is the provision authorizing covers that “imitate or simulate those in the copyrighted sound recording.” The Office appears to be suggesting that cover songs created by real performers should still be allowed but not if they are produced by digital replicas. This is an appropriate and important distinction.
Rejects Calls for “Full” Preemption of State Law
The Office “recommends against preempting state law.” I agree with the Office that it is not wise to preempt 100s of years of state laws that address unauthorized uses of people’s voices and likenesses. However, I disagree that a federal digital replica law (or broader publicity law) should not preempt some state laws at least in part. Federal laws, notably including copyright, have been able to incorporate longstanding common law rules even as they set forth a national standard. Failing to preempt the burgeoning set of state digital replica laws will greatly worsen the “thicket” of conflicting laws that address unauthorized uses of a person’s identity, including their voices and likenesses. Failing to preempt state laws will make identity rights increasingly difficult to navigate for artists, creators, and other speakers. See Jennifer E. Rothman, Navigating the Identity Thicket, 135 Harvard L. Rev. 1271 (2022).
Importantly, not preempting state laws (at least in part) misses an opportunity to protect individuals from state laws that troublingly claim to allow someone to own another person’s name, likeness, and voice.
Some Objections to the Report's Statement of the Law
I want to reiterate that the Copyright Office did an impressive job navigating challenging political waters as it put together its report. There is a lot to agree with in the Report’s recommendations, though I have some areas of disagreement that I have highlighted above. As a scholar and expert on right of publicity law, I also cannot help pointing out some additional areas in which I disagree with the description of the law. In some instances, these disagreements may affect the urgency and need for federal legislation, in others it is a matter of correcting the historical record—a challenging endeavor given that some very wrong things have been said about the right of publicity for a very long time and then been repeated over and over again. I would be remiss if I didn’t take this opportunity to correct some of these inaccuracies.
The Right of Publicity did NOT appear in 1953. It dates to the 1800s. Sadly, the Office repeats the frequently said, but erroneous, claim that the right of publicity was “coined” by the Second Circuit in the Haelan Labs v. Topps Chewing Gum case in 1953. Both the term and the claim long predated the decision, and the case itself did not involve a right of publicity claimant and was in reality a contract claim. I extensively debunk the mythology surrounding Haelan in my book on The Right of Publicity. (Jennifer E. Rothman, The Right of Publicity: Privacy Reimagined for a Public World (Harvard Univ. Press 2018)). For this section of the book, I read the trial court and appellate court documents in the case and tracked the law both before and after this case through significant archival work. I previously noted this error in the Office’s 2019 report on moral rights, but it’s hard to fault the Office given that this error is repeated so often in scholarship, the dominant McCarthy treatise, and perhaps being reified even as I write this by AI and its likely results when answering questions about the right of publicity.
I point this out not only because I find it frustrating but also because it shifts how we think about the right. For those who think the right of publicity was something new and different from the appropriation branch of the privacy, they often think the right of publicity is a commercial right for the famous that doesn’t encompass noneconomic injuries and that privacy-based appropriation claims are only for those who shy away from the public eye. Both conclusions are wrong historically and as a doctrinal matter today. And they also are highly problematic substantively.
Thankfully, the Copyright Office recognizes that on the ground state law protects people’s names, likenesses, and voices through both privacy and publicity laws that are often interchangeable with one another, and that normatively we should want to protect the famous and the ordinary alike.
The Right of Publicity Applies to Noncommercial Uses. The confusion over when the right of publicity arose and why can also be seen in the Report’s mistaken claim that state publicity laws “typically apply only …in advertising, on merchandise, or for other commercial purposes.” This is largely not true, particularly with regard to claims by living people. Right of publicity claims have applied to uses in news, video games, movies, magazine stories, and works of art. See Jennifer E. Rothman, Commercial Speech, Commercial Use, and the Intellectual Property Quagmire, 101 Virg. L. Rev. 1929, 1953-55 (2015). In fact, the only right of publicity case heard by the U.S. Supreme Court involved a nightly news broadcast that showed a clip of a performer at a local fair. See Zacchini v. Scripps-Howard Broadcasting, 433. U.S. 562 (U.S. 1977).
The Office is absolutely right, however, that the harms from uses of digital replicas can occur even when users does not seek “financial gain.” It is also correct that distinguishing between commercial and noncommercial uses is a challenging matter. (The Report cites to Stacey M. Lantagne, Famous on the Internet: The Spectrum of Internet Memes and the Legal Challenge of Evolving Methods of Communication, 52 U. Rich. L. Rev. 387, 416-17 (2018); see also Rothman, Commercial Speech, Commercial Uses, at 1959-65, 1974-77).
I applaud the Office’s call for any federal digital replica law to encompass both “non-commercial and commercial uses.” So the recommendations are right, even if I can quibble with the Report’s description of the state of the current law or the history of its development.
* * *
In sum, the Office’s Report will be helpful for Congress and others as we think through how best to address this age of AI and the dangers we face as a result of digital replicas. The Office’s approach is understandably filtered through its expertise in copyright law and its focus on creators and the creative industries. Such a perspective represents an important one in thinking through these issues, but it is not the only one given the impact of digital replicas on individuals’ autonomy, livelihood, and dignity, as well as the danger of deepfakes to deceive the public.