The music industry has welcomed a revised version of the NO FAKES Act which was introduced in Washington yesterday. This is the act that will create a new US-wide right to help performers and creators, and people more generally, control their voice and likeness in the context of AI.
Although generally supported by the music industry from the off, there have been critics of the act and the new ‘digital replica right’ it will introduce, including from some media companies and digital platforms, such as Spotify.
Supporters of the act hope that the latest amendments will address concerns expressed by those critics. For its part, Spotify says there is “still work to do”, but that it “appreciates” the improvements made by Marsha Blackburn and Chris Coons, the two main senators behind the proposals.
Launching the revised act, Coons says the new version is “stronger” after “working with stakeholders from across the country to make the protections in our bill more robust without compromising Americans’ free speech rights”. He then adds, “Whether they’re Tom Hanks or an eighth grader in Wilmington, no one should worry about someone stealing their voice or likeness”.
Meanwhile Blackburn adds “from artists and songwriters to students and everyday Americans, people deserve meaningful protections against deceptive deepfakes and digital impersonation”.
Speaking for the US record industry, RIAA CEO Mitch Glazier cites polling that says 92% of Americans “are concerned about the impact of AI deepfakes on our neighbours and culture” and which “reveals near total support for a federal law that protects voice and likeness”.
The NO FAKES Act, he adds, “provides those important protections while securing freedom of expression, reducing litigation and achieving the full promise of American AI technology”.
The NO FAKES Act was originally proposed in 2023, and then introduced in US Congress in 2024 and reintroduced last year. It was known that music streaming services, especially Spotify, had concerns about how the new digital replica right would work in practice, and what obligations and liabilities would fall onto digital platforms as creators and promoters sought to enforce the new right.
The new version of the act includes more flexibility regarding how enforcement will work depending on the kind of platform where AI-generated voice clones and deepfakes are being published. There is also a ‘counter-notice’ system for challenging the removal of content under the act, and new exemptions to benefit libraries, archives and research institutions.
Commenting on the new version of the act this week, Spotify says, “while there is still work to do, we appreciate Senators Blackburn and Coons’ work to improve the NO FAKES Act on the whole, which puts meaningful guardrails in place for the AI era”.
Graham Davies, CEO of DiMA - the US trade body for music streaming services - says the latest version of the act includes “improvements” that “help provide certainty and clarity for music streaming services”. While there are “outstanding issues”, he adds, “we are encouraged by the progress made” and “look forward” to “ongoing dialogue as the bill advances through the legislative process”.
Regulating digital replicas
As unapproved AI-generated voice clones and deepfakes have become widespread, there has been much debate about how creators and performers can control their voice and likeness in the AI domain.
For an AI to imitate a person’s voice or likeness, it likely needs to be ‘trained’ on existing audio and/or visual content featuring that person. That involves copying that content and in theory the AI company should get permission from the relevant copyright owners, who can always decline to give permission.
However, many AI companies argue they don’t need permission from copyright owners because AI training is ‘fair use’ under US copyright law. But even if they do need rightsholder consent, creators and performers don’t necessarily own the copyrights in their work.
In the US there is another principle called ‘publicity rights’ which allows creators and performers to control the use of their image or identity, and this can probably help when it comes to digital replicas.
However, publicity rights come from state-level laws and therefore differ around the country. The NO FAKES Act will introduce a US-wide system.
The UK dimension
In the UK, publicity rights don’t currently exist in law. This was raised in a House Of Lords report on AI earlier this year, which said “the government should introduce protections against unauthorised digital replicas”.
Those protections, the report said, should provide creators and performers “clear, enforceable control over the commercial exploitation of their identity” while also “appropriately safeguarding freedom of expression and other legitimate uses”.
The UK government responded to that report last week. It noted that – under UK law – the legal principle of ‘passing off’ could help stop unapproved digital replicas.
It then described that principle as “a common law tort intended to prevent businesses misrepresenting their goods or services as the goods or services of another business”.
However, the government’s response conceded that “this avenue is less accessible for lesser-known artists or the general public”, while protections from copyright or performer rights aren’t necessarily that helpful either.
Ministers previously announced that they would be doing more work on this issue and, in their response to the House Of Lords, they said, “the government will launch a consultation on digital replicas in the summer”.
That “will seek views on how to address the harms caused when someone’s likeness is replicated without their permission, while protecting legitimate innovation”.