Deepfakes, digital doubles, and the law: Jennifer Rothman on protecting identity in the AI era

Artificial intelligence technology has made it possible to produce digital replicas of real people—deepfakes, voice clones, and other once-unthinkable products—making it harder than ever to tell what is real, says Penn Carey Law professor Jennifer Rothman.

This leap, she says, means the law needs to more robustly protect the people at the center of such unauthorized uses of a person’s identity, while cautioning that existing laws may need to be more vigorously enforced instead of putting new ones in place.

“We need to focus on protecting each of our rights to control uses of our own voices and likenesses, rather than solely focusing on how we harness as much market value from these rights as possible,” says Rothman, a global expert in intellectual property law and the Nicholas F. Gallicchio Professor of Law, with a secondary appointment at Penn’s Annenberg School for Communication. Her work has recently focused on how the law should respond to the rise of artificial intelligence.

She gave the Donald C. Brace Lecture for the U.S. Copyright Society with the title “Copyrighting People,” published recently, raising questions about the intersection of U.S. copyright law and the right of publicity. The latter is a type of state law that protects against unauthorized uses of a person’s name, likeness, or voice.

Rothman warns against copyright law becoming a vehicle for others to control a person’s identity, whether the person captured in the work is a performer, politician, athlete, or an average citizen.

In the enthusiasm to protect against deepfakes and other troubling uses of AI technology, “we shouldn’t create a system that will vest control of people’s digital replicas in anyone other than the person themselves,” says Rothman, a leading expert on the right of publicity and personality rights who is frequently turned to for her expertise on state and federal legislation. “The concern there is that from an early age people will essentially be signing over their own—or their kids’—rights to their digital replicas and voices, perhaps unwittingly, or perhaps because they think it’s a way to get some cash.”

The idea of excluding people from copyright protection is “increasingly unstable,” Rothman says. Copyright has its roots in the U.S. Constitution, but at the time it was written photographs, videos, and computer software—all protected by copyright law now—didn’t exist.

Today, she says, there is an open question of whether digital replicas of people will be considered copyrightable, and, if they are, how this will impact the person’s control over how such a replica is used. There are also questions about whether the person would infringe these replica rights by creating new performances of their own.

Even outside the copyright system, Rothman says, new state laws and proposed bills at the federal level create a quasi-copyright system for digital replicas and voice clones, and do so in ways that allow someone other than the person depicted to control the use of that person’s own digital replica and voice.

Rothman paints a possible dystopian future of not only technology that can deceptively make it appear as if we are saying and doing things we never did but also a legal regime which legitimizes such a possibility. She cautions against rushing through legislation which may worsen the circulation of deepfakes and lack of personal control over digital replicas.

“Many laws on the books already cover a lot of this, so it’s not clear where we need intervention yet,” Rothman says. “A carefully crafted federal law could help standardize the wild west of digital replica and AI laws currently being adopted at the state level. But Congress and states should do so only if the legislation robustly protects our control over our digital selves, as well as the public from deception.”