It started with a song nobody made.
In April 2023, a track called Heart on My Sleeve appeared on Spotify, Apple Music, and TikTok. It had the unmistakable low rumble of Drake and the wounded falsetto of The Weeknd. It was slick, moody, radio-ready — and entirely fabricated. No studio session had taken place. No contract had been signed. A producer tag from Metro Boomin floated over the intro, also synthesized without his knowledge or consent. Within days, the song had racked up millions of streams. Universal Music Group eventually killed it with a DMCA takedown, but the damage — or rather, the demonstration — had already been done. The technology to clone a human voice had become cheap, fast, and absurdly accessible to anyone with a laptop and a grudge or a dream.
What came after that moment has been one of the stranger legislative sagas in recent memory: a three-year slog through Senate subcommittees, a Hollywood strike, a country music legend’s medical crisis repurposed into an argument for federal law, and — as of this week — a third attempt to pass a bill called the No Fakes Act before time runs out on another Congress.
The Gap Nobody Wanted to Talk About
To understand why this bill keeps coming back, you have to understand the legal vacuum it’s trying to fill.
Copyright law protects original works — scripts, scores, finished films. Trademark protects names and logos in commerce. But there has never been, at the federal level, a law that says a person owns their face or their voice. That right — called the “right of publicity” — has existed only in a patchwork of state statutes, roughly 35 of them, varying wildly in scope and enforcement. What that means in practice is that an AI company operating in one state might enjoy protections unavailable in another, while a performer in a third state has essentially no recourse when her voice shows up in a product she never agreed to endorse.
The No Fakes Act — its full name is the Nurture Originals, Foster Art, and Keep Entertainment Safe Act, an acronym that required what must have been a particularly committed legislative staffer — was conceived to plug that hole. It would establish, for the first time, a federal intellectual property right tied to a person’s voice and visual likeness. Unauthorized use of someone’s digital replica without their consent would be actionable. And the right would not die with its holder: under the bill’s current terms, heirs and estate executors could license and transfer it for up to 70 years after death, consistent with the term lengths of U.S. and EU copyright law.
The bill was first floated as a discussion draft in 2023. It was formally introduced to the Senate in the summer of 2024. It died in committee at the end of the 118th Congress without a vote. It was reintroduced in April 2025, with considerable fanfare, by a bipartisan group of senators — Marsha Blackburn (R-TN), Chris Coons (D-DE), Thom Tillis (R-NC), and Amy Klobuchar (D-MN) — headlined by a press conference where, extraordinarily, YouTube, Amazon, and OpenAI all showed up to endorse it. It died again. And on May 20, 2026, yet another revised version landed in both chambers, this time with Spotify and Getty Images newly on board, and with a coalition so broad it encompasses the three major music labels, the RIAA, the Recording Academy, SAG-AFTRA, the Motion Picture Association, and a roster of AI companies whose business models would, in a different world, make them the natural opposition.
That last fact — tech giants lining up to support restrictions on themselves — says something about where the culture war over artificial intelligence has quietly ended up.
The Strike That Changed Everything
To understand why the entertainment industry has been so galvanized on this issue, you have to go back further than Heart on My Sleeve. You have to go back to July 2023 and the beginning of the SAG-AFTRA strike — the most significant Hollywood labor shutdown in more than six decades.
The strike was about residuals, streaming revenue, and the fundamental restructuring of how entertainment gets made. But as negotiations dragged on through a brutally hot Los Angeles summer and into the fall, one issue kept surfacing with particular urgency: AI. Specifically, the fear that studios could use digital technology to scan an actor’s face and body on the first day of employment, pay them a day’s rate, and then replicate their performance indefinitely — for free, forever, without further consent or compensation.
Fran Drescher, then president of SAG-AFTRA, called it at the union’s founding strike announcement what it was: an existential threat. Not a future worry. A present danger. “When the evildoers are actually trying to put words into my mouth that I never said,” she said later, “that goes against my moral compass and everything I stand for.” Background actors reported that body scans were already being stored for potential AI reuse — described in some negotiations, chillingly, as being used “for the rest of eternity” — without clear terms or additional pay.
The strike lasted 118 days. When it ended in November 2023, the contract secured meaningful protections: consent requirements for digital replicas, obligations to notify the union when synthetic performers were deployed in place of human ones. But the union made no secret of what the deal was: a stopgap. The real protections would have to come from Washington.
Then, eight months after the strike ended, Scarlett Johansson publicly confronted OpenAI. The company had released a voice assistant called Sky that sounded, with unnerving precision, like Johansson’s character in the 2013 film Her — the same role the company’s CEO Sam Altman had explicitly cited before the launch. Johansson said she had declined twice to license her voice to OpenAI. The company pulled Sky within days, but the episode landed exactly as it was bound to: as proof that even the most powerful performers in the world couldn’t protect their own identities through private negotiation alone. There was no federal law to sue under. There was no floor.
Randy Travis at the Podium
In April 2025, when lawmakers formally reintroduced the bill for the second time, they didn’t bring a lawyer to the press conference. They brought Randy Travis.
Travis, the country music legend who suffered a massive stroke in 2013, lost the ability to speak and sing the way he once had due to aphasia — a condition that affects language processing after brain injury. He hadn’t released new music in a decade. But in 2024, he recorded a new song, Where That Came From, using AI to reconstruct his voice from thousands of hours of pre-stroke recordings. The result was released commercially and went to number one on the country charts. It was, depending on who you asked, either a moving triumph of technology in service of human expression, or a preview of exactly what the No Fakes Act was designed to prevent — except that Travis had consented to every step of it.
His presence at the Capitol was strategic and pointed. Speaking through his wife Mary, Travis drew the line precisely: “These are very different from someone else stealing my voice and producing music, music that I never participated in or authorized… no one should be allowed to put words in someone else’s mouth or depict them doing something they never did.” The bill’s supporters needed that distinction to be legible to legislators who might worry about overcorrecting — outlawing beneficial, consensual AI applications while trying to stop the bad ones. Travis was the argument made flesh.
The 2025 reintroduction came with an unusual coalition. Tech companies that have spent years lobbying against content restrictions signed on. YouTube’s VP of Public Policy, Leslie Miller, said the company had worked with the RIAA and the Motion Picture Association to push for a “shared consensus” on the legislation, citing nearly two decades of experience in rights management at scale. OpenAI, IBM, Adobe, and Amazon signed their names. Stephen Fry — the British actor who had revealed in 2023 that his voice had been scraped from his Harry Potter audiobook recordings and replicated without permission — became an informal transatlantic avatar for the cause.
In April 2026, months before this week’s latest reintroduction, Senators Blackburn and Peter Welch held a roundtable with more than 20 artists during the Recording Academy’s GRAMMYs on the Hill Advocacy Day. Spotify and Getty Images — a company whose entire business model depends on the controlled licensing of images of real people — joined the coalition. The bill that landed in Congress on May 20 now carries the backing of an alliance that would have seemed implausible three years ago: Hollywood unions, Nashville songwriters, Silicon Valley product teams, stock photo agencies, and a bipartisan Senate duo that agrees on almost nothing else.
What the Bill Actually Does — And What It Doesn’t
The No Fakes Act, in its 2026 version, would give every individual in the United States the right to authorize — or refuse — the use of their voice and visual likeness in digital replicas. It imposes liability on anyone who publicly displays, distributes, or uses an unauthorized digital replica. It includes a notice-and-takedown mechanism modeled loosely on the DMCA, allowing victims to demand removal from platforms. Online platforms that adopt anti-violation policies and respond promptly to takedown requests get safe harbor protection.
The bill already had carve-outs for news reporting, documentary work, sports coverage, biographical works, commentary, criticism, and parody. The 2026 revision adds a counter-notice procedure — allowing creators to challenge takedown demands — and an exemption for certain work at libraries, archives, and educational institutions, a concession to the academic community that had raised First Amendment concerns about the earlier drafts.
Those concerns haven’t gone away. Legal scholars have noted that the preemption provision — the clause that would override state publicity laws — is ambiguous in ways that could trigger years of costly litigation. The question of what counts as a “digital replica” under existing state law versus the bill’s new federal framework is genuinely murky. And critics have pointed out that the bill’s First Amendment exposure is real: any federal law regulating the reproduction of a person’s image and voice in expressive works will attract constitutional challenges from the moment it’s signed.
SAG-AFTRA has also escalated its own battles in the meantime, filing an unfair labor practice charge in 2025 against Llama Productions over the use of an AI-generated version of James Earl Jones’ voice for the Darth Vader character in the video game Fortnite. The union argued that replicating a deceased performer’s voice without bargaining not only violated member rights but deprived living human performers of potential work. The case threw into sharp relief what the bill’s post-mortem provisions are really about: not just sentiment, but jobs, and who does them.
The Longer War
The No Fakes Act exists within a larger legislative landscape that keeps moving faster than any single bill can track. The TAKE IT DOWN Act — targeting non-consensual intimate imagery, including AI-generated deepfake pornography — passed. The DEFIANCE Act, giving victims of sexual deepfakes a federal civil cause of action, was reintroduced in 2025 after expiring at the end of the previous Congress. The Protect Elections from Deceptive AI Act, banning AI-generated deepfakes in federal political advertising, has been floating around since 2023 without resolution. State-level laws have multiplied, creating exactly the jurisdictional chaos the No Fakes Act is meant to supersede.
Congress has a pattern with technology legislation: it moves slowly, often too slowly, until some galvanizing public event turns a bureaucratic process into a political imperative. The DMCA took years and a crisis. COPPA took years. GDPR wasn’t American at all — it was European pressure that eventually forced the American conversation. The No Fakes Act has now survived three Congresses, two rounds of industry coalition-building, and one legislative death by committee. Whether the third time produces a floor vote depends, as it usually does, on factors that have nothing to do with the merits of the bill.
But something has shifted in the politics of this. The coalition that showed up in 2025 and showed up again in 2026 is not the coalition that usually loses in Washington. It is not a narrow interest group making narrow arguments. It is an alliance of the entertainment industry, the technology industry, the creative labor movement, and a bipartisan Senate roster all pointing at the same problem. The RIAA’s chairman Mitch Glazier, whose organization has spent decades in adversarial litigation with tech platforms over exactly these questions, framed it plainly: “Every individual should have autonomy over these fundamental personal qualities, and the NO FAKES Act is the first step — it’s your voice on the line.”
Universal Music Group, which was the entity that killed Heart on My Sleeve via DMCA three years ago — because that was the only available tool — put out a statement applauding the reintroduction and calling the bill “landmark, bipartisan, bicameral legislation to address deepfakes and other threats to individuals’ rights to control their own voice and visual likeness.”
The song that started this, at least in part, is gone. But the question it raised isn’t: in an era when a voice can be reproduced with a few clicks and a prompt, and when the law has no clear answer about who owns it, what does it mean to be the original?
The No Fakes Act is, at its core, a bill about that question. It’s a bill about whether the most intimate thing a person has — the particular frequency of their speech, the specific geometry of their face — belongs to them, or to whoever figures out how to copy it first.
Three years in, it still hasn’t passed. But the room it’s walking into keeps getting bigger.
Enjoyed This Story? Subscribe to EntertainLens for more entertainment insights.
