Three years ago, Tom Hanks posted a warning to his millions of fans: “BEWARE!!” A dental plan was running an A.I.-generated ad featuring a younger version of the actor — and he had nothing to do with it.
It’s not just celebrities who are being impersonated by A.I. Voice actors who were paid a few hundred bucks for recordings they were told were for internal research later found clones of their voices for sale online. A content creator learned on her honeymoon that a YouTube ad was using an A.I. version of her to sell erectile dysfunction supplements.
Is all this legal? Right now, it’s an open question. But it shouldn’t be.
Hollywood doesn’t need America’s leaders to devise a new concept for new circumstances. We need them to extend and enforce the rules that already exist. The use of Americans’ faces, voices, performances and creative works is already subject to rules at the federal or state level that say, in short: You need to get permission, and you need to pay for the use.
This principle has guided the entertainment industry for more than a century. Whether it can continue to protect Americans may be decided in the coming weeks. The No Fakes Act would create a federal prohibition on creating nonconsensual digital replicas of any person’s voice or likeness — famous or not. It was approved unanimously by the Senate Judiciary Committee in June. There is momentum to pass it through the full Congress before the end of the year. It’s one of the few A.I. bills with true bipartisan support.
The Senate and House should move quickly and deliver the bill to President Trump for signature. That would ensure Americans retain control of their images and voices in the A.I. era.
I come to this issue not as a policymaker but as a practitioner. I run a global talent agency. Part of our job is to help our clients license their likenesses and performances, sometimes to A.I. companies. We regularly negotiate those deals, and I can tell you that the bill pending in Congress delivers what the industry needs and what will protect all Americans. Some A.I. companies already follow many of its measures voluntarily.
Licensing is not overly complicated. While there are many profound — even existential — questions raised by A.I., this is among the easier ones. When Creative Artists Agency clients like Scarlett Johansson, Cate Blanchett or George Clooney appear in an ad, there is a contract and a payment. Radio stations pay to license the songs they play. When Steven Spielberg makes a movie, no one tells him it’s too complicated to clear the rights for every location depicted, every product featured, every piece of music heard.
These legal principles are almost as old as the country. In 1790, George Washington signed the first federal copyright law, covering maps, charts and books, and Congress has extended it to almost every new medium since — photographs in 1865, motion pictures in 1912, sound recordings in 1971. With every new technological wave, the rules of copyright, contract law and licensing adapt to accommodate it, and A.I. should be no different.
The No Fakes Act would apply a tried-and-true formula to the A.I. era: control, consent and compensation. It would set a uniform federal standard to replace outdated, ill-fitting state laws, many of them written with an analog world in mind. Its supporters include all sides of the table: SAG-AFTRA, the union representing actors and other performers; the Motion Picture Association, representing the studios; and tech companies including IBM, Google and OpenAI.
This coalition isn’t asking Congress to regulate A.I. broadly. It’s not calling for onerous taxes or compulsory licensing schemes or for the ability to control anyone else’s content. The bill protects news, commentary, scholarship, satire and parody, reflecting the same free-expression interests long recognized in copyright law. We’re simply asking Congress to extend the principles that have defined our industry for a century to a new technology.
Hollywood is not afraid of new technology. The objection from actors, musicians and other creators is to a world in which anyone can use their face or voice without permission. In fact, many artists are already embracing A.I. on their own terms. Some of them are making deals with companies that are creating their A.I. likenesses, with their consent. The A.I. companies are paying for use, and not only in Hollywood. OpenAI signed a five-year licensing agreement with News Corp, the publisher of The Wall Street Journal, reportedly worth more than $250 million.
Without clear laws, harms would fall hardest on everyday creators — emerging writers, artists and musicians — who deserve just as much protection as their more established peers. Their work is just as easy to clone as that of a celebrity, and they often lack the resources to fight back. I also worry about the effect on individuals who may not even know their image is being used without their consent.
The company behind the fake Tom Hanks ad could have gotten permission to license his likeness. Would he have wanted to be pitching a “miracle” tooth treatment? Probably not. Every one of us should have the right to say no to an unwelcome use of our likeness. What’s missing is a bright-line rule affirming that right in the A.I. age, and it is one Congress can supply before the year is out.