FKA twigs built a digital twin of herself, then told the Senate artists must control it

On 30 April 2024 FKA twigs sat in front of the Senate Judiciary Subcommittee on Intellectual Property and told senators she had made a deepfake of herself. Over the past year, she said, she had developed a version of herself "not only trained in my personality but can also use my exact tone of voice to speak many languages", according to Tech Policy Press's transcript of the hearing. Then came the sentence the rest of her testimony hangs on: "This, however, is all under my control and I can grant or refuse consent in a way that is meaningful."

The hearing was on the NO FAKES Act, still a discussion draft. Subcommittee chair Chris Coons said he had released it the previous October with Senators Tillis, Blackburn and Klobuchar. PBS NewsHour streamed it live. She was one of six witnesses, and the only working artist. Her label boss, Warner Music Group CEO Robert Kyncl, was another.

What is her AI twin for?

Her written statement, published before the hearing by Stereogum via MusicAlly, says she would be "engaging my AI twigs later this year to extend my reach and handle my online social media interactions", while she kept working "from the comfort and solace of my studio". MusicTech reported the same plan. So on the day, the twin was a tool in development, not one already talking to fans.

Answering Coons, she was more specific. She said she had "currently explored French, Korean and Japanese", so she could explain her upcoming album to fans in their own language. She also named the chore it takes off her plate: being an artist now "requires a lot of press and a lot of promo, a lot of one-liners". For something simple that doesn't need her heart, the twin can deliver the line, and she gets the time back for the work.

That's a narrow, specific use: translation and repetitive promo, not new songs or performances. It's also the strongest version of her argument. She isn't asking senators to stop the technology. "Let me be clear. I'm not against AI," she said in her opening statement.

Where did she draw the line?

At consent. What isn't acceptable, she said, is when "my art and my identity can simply be taken by a third party and exploited falsely for their own gain without my consent". She described songs online, including collaborations with other artists, that she never made, and the risk that someone could "change lyrics, change messaging" in her voice. She also said 85 of her songs had leaked in the previous six months. Her label got them taken down, but a fake that says something harmful in her voice, she argued, can't be taken back once people believe it.

She framed what she wanted as control. She said she spoke for creators who need "tight control over their own art, image, voice, and identity", and later: "I want to be in control of my likeness, my brand, my legacy." Karen Smiley's newsletter, summarising BBC coverage, described the testimony as being about artists controlling their "name, image, likeness, and voices, as well as their content" (her summary). Her own words, in the transcript, were art, image, voice, identity, likeness, brand and legacy.

What did the others at the table want?

Coons said the draft would hold people and companies liable for making an unauthorised digital replica of someone's voice, image or likeness, and platforms liable if they host one knowing it wasn't authorised. Unlike many state right-of-publicity laws, which he said often focus on celebrities, it would cover everyone, not just people who commercialise their likeness.

Kyncl made the economic case. Fake songs in an artist's voice, he said, compete with her real ones for a fixed pool of streaming revenue. He wanted a right each person can "license or deny" on free-market terms. He also told Senator Tillis that record deals increasingly include likeness rights, so labels can file takedowns for their artists. Asked whether the new right should be fully transferable, he said the artist "should have a choice to either transfer or license".

SAG-AFTRA's Duncan Crabtree-Ireland pushed the other way on that point. He said a young artist shouldn't be able to sign away their image, likeness or voice for 50 or 60 years with no way to renegotiate, and suggested something like a seven-year cap on transfers. Graham Davies of the Digital Media Association, which represents streaming services, wanted one federal law that preempts state rules, because a state-by-state patchwork is hard for a global business.

Twigs, asked by Tillis, said the power should sit with artists and with "people who are there to protect the artists, whether that's third parties like record labels or agents or lawyers". "That's up to the artist to understand and to sign a contract if we want to," she added.

What we think

Her testimony is the best case yet that "AI or no AI" is the wrong question for artists. Her twin and a scammer's deepfake use the same technology. What separates them is who decides what the voice says and who keeps the money. She drew that line clearly, and her use case, multilingual fan contact and promo filler, is exactly the kind of drudgery many artists would happily hand off.

The gap is in the room's other testimony. Control that can be transferred in a contract is only as strong as the artist's bargaining power on the day they sign. Kyncl, her own label's CEO, argued for transferability and said likeness rights are already showing up in record deals. A star with leverage can hold on to her twin. A 21-year-old signing a first deal, the example Crabtree-Ireland gave, may not get to choose. If the final bill lets the right be transferred without a short time limit, it risks moving control of the digital twin from the scammer to the contract.

So the details to watch aren't the headline right. They're whether the right can be transferred or only licensed, how long a licence can last, and whether a licence has to name specific uses. An artist's AI twin is worth having only if the artist can still switch it off.

A note on our sources: we quote twigs from the Tech Policy Press transcript, which it calls lightly edited, and from her written statement as published by Stereogum. The committee's hearing page holds the official record but blocked our fetcher, so we haven't checked these quotes against it.

Sources

• Tech Policy Press, Transcript: US Senate Judiciary Subcommittee Hearing on "The NO FAKES ACT" (1 May 2024)
• US Senate Committee on the Judiciary, The NO FAKES Act: Protecting Americans from Unauthorized Digital Replicas (hearing page, 30 Apr 2024)
• Stereogum, FKA Twigs Developing Deepfake To Talk To Fans, Testifies Before Senate Committee About AI Regulation (30 Apr 2024)
• MusicTech, FKA twigs is developing an AI version of herself to outsource fan interactions (30 Apr 2024)
• PBS NewsHour, WATCH: Senate committee holds hearing on unauthorized digital replicas with singer FKA Twigs (30 Apr 2024)
• 6 'P's in AI Pods, Musician-led AI initiatives (10 Jul 2024)