Current Digital Replicas Legislation – Misguided Proposals Miss The Mark
Do you recall watching Star Wars: Rogue One and seeing a posthumous Carrie Fisher’s Princess Leia or Peter Cushing’s Grand Moff Tarkin? Or what about Tupac performing with Snoop Dogg and Dr. Dre at Coachella 2012, sixteen years after his passing. Using everything from cutting-edge technologies to simple Photoshop, so-called “digital replicas” are produced every day. Some are satirical, some biographical, while others are used entirely for educational purposes. Despite digital replicas existing for decades, recent discussions surrounding celebrity uses such as Taylor Swift and Scarlett Johansson have vaulted the subject to the halls of Congress. Yet, in seeking to protect our digital likenesses, we may need to take a step back to study existing constitutional and copyright laws that have been on the books for hundreds of years.
Numerous legislative proposals — such as NO FAKES, No AI FRAUD, and Preventing Abuse of Digital Replicas Act — have emerged with the intent to protect the voice and likenesses of individuals. While laudable, these pieces of legislation may complicate existing legal protections by lumping many categories of digital replicas into one theory of harm, without appreciating the nuances between cases. For example, a duty to limit non-consensual intimate images is entirely different than a dispute over Harry Potter fanfiction.
Even more confusing, defining a digital replica seems to be a subject of debate. Many current proposals include varying definitions, muddying the water as to what rests in that category. Common among the definitions is that digital replicas are computer-generated, highly realistic simulations of a person’s (living or not) image, voice, or likeness. These simulations are then used to create new images, videos, or audio of that person.
Utilizing technological reproductions is not solely the enterprise of film studios attempting to put famous faces in their newest blockbusters. Thanks to this technology, Randy Travis, a famous country music singer and songwriter, who lost his voice after a tragic stroke in 2013 led to aphasia, created new music this year. Travis’ wife, Mary, credits AI, calling it “artistic intelligence” now that he can make music again. Last year, the Beatles produced a final record by sampling a long forgotten demo of a song John Lennon wrote, expanding on Lennon’s vocals to create the song “Now and Then.” Paul McCartney and Ringo Starr utilized AI to clean up Lennon’s audio from that demo track, improving the clarity of his voice to make the new song. Other technologies are being deployed in healthcare, speech, creative endeavors, art, and more.
Are these examples all considered digital replicas? While unclear, an illustration of the difficulty of drawing a legal line emerges. After all, many legislative proposals impose new rules on both the individuals who create the digital replica, and on platforms that simply host the content.
In a world where these lines are blurred, tools that are utilized by individuals every day like Photoshop or Auto-Tune would now create serious questions of liability. And at what point in the improvement or alteration of a work would the line be crossed between a simple production and an unlawful digital replica? To go even one step further, since many famous songs use samples from other hits, who ultimately is the legal holder of a digital replica right and at what moment does ownership attach?
These questions, combined with implications of emerging technologies that are undoubtedly far-reaching, foster novel debates around commercial use, personal privacy, and more. Luckily, existing laws can help answer many of these questions. Section 230 allows digital creators to thrive and create online, the Digital Millennium Copyright Act protects copyright holders from bad actors who seek to unlawfully profit off others’ works, and the First Amendment of the Constitution provides the backstop for protecting speech as one of our nation’s highest ideals.
New legislative proposals seek to add yet another layer, this time cabining concerns such as electioneering, privacy, and non-consensual intimate imagery into a single theory of harm. Without appreciating that these problems each require different tools, proposing a single universal solution could create more complications than solutions, akin to using a hammer for a nail, screw, and buzzsaw.
First, many states have already enacted right of publicity laws. Besides creating confusion between a federal and state framework, inconsistent definitions, rights, and remedies may create instances where it is impossible to comply with all existing legal obligations. These proposals could additionally impact the functioning of the internet by placing liability not only on the parties responsible for the creation of the digital replicas, but also on the digital services that host the content. For instance, the NO FAKES Act would attach liability to any digital service where a digital replica is uploaded. The potential for unbound liability could lead to businesses needing to be intimately aware of everything that they host. More likely, many businesses would simply remove, or refuse to publish content, to avoid liability. This result would discourage innovation and silence artistic creativity.
As digital replicas advance, it is understandable that Congress would start to contemplate protections against the abuse of one’s digital likeness. However, legislators should first consider the current slate of rights and remedies in place and the application of technologies. New legislation, if appropriate, should consider solving identifiable loopholes rather than attempting to craft one silver bullet which would apply to any conceivable harm. There are undoubtedly a number of delicate issues present. But we must not pursue a regime that unintentionally causes more harm than good.
