(Almost) anything goes for AI in podcasting

I am working on a segment for The Vergecast on the use of AI in podcasting, and I thought I would give you all a little preview of my conversation with Emily Poler, a commercial litigator who specializes in issues of intellectual property and technology.

I first spoke with Poler last fall, and since then, there have been a number of developments indicating how AI may be managed within podcasting. In the absence of policy explicitly dealing with the subject, companies have made little pivots and rules hemming in AI use. Earlier this month, The Ringer’s union eked out a win by requiring Spotify to get consent from staffers to clone their voices — except in cases of translation. Meanwhile, Apple Podcasts and YouTube now require creators to disclose the use of AI in their work. Tennessee, one of the major centers of the music industry in the US, passed a law aimed at protecting music artists from having their voices cloned.

The rules and norms around AI are rapidly evolving. Below, I talk to Poler about how podcasters can use AI in their shows and what practices could get them into trouble.

This interview has been condensed and edited for clarity.

Say someone’s making a podcast and they want to clone the voice of a famous person. Maybe they have editorial reasons for this — they could want to do a voiceover for what’s a written social media post — or they have some other need to incorporate that. Are there any legal restrictions preventing them from doing so?

There are two avenues of legal attack there. One is the right of publicity statutes that limit the ability of another person to use, it can be anybody’s or it can be a celebrity’s, image, voice, or likeness in some capacity. That’s one area. And the other area is going to be under both federal and state unfair competition law. The federal law is called the Lanham Act, and there are prohibitions creating a situation where it seems like a celebrity is endorsing a product or is affiliated with the product.

There could be situations where, if you used what appears to be a celebrity’s voice in connection with a podcast — for example, you create an AI-generated clip of what appears to be somebody reading their X, Threads, or Instagram post — the issue that’s going to come up is whether it appears that that person cooperated or participated in the making of the podcast. And I think that’s where you’re going to see potential problems.

One other thing to keep in mind is that there’s a pretty significant First Amendment issue underlying all of these laws, particularly on the right of publicity. I think in every state law where there is a right of publicity — and about two-thirds of states have a right of publicity law of some shape, manner, or form — those laws only apply in the commercial context. The next question is, “Well, what’s commercial?” and it’s a pretty common understanding that commercial means the podcast has sponsors, and therefore, it’s commercial. No. That’s not what commercial means in this context. Commercial means that it is literally an advertisement. So you can’t have Britney Spears marketing your local Kroger or ShopRite if she has not, in fact, consented to do that. The New York Times has advertisements and makes money and is quote, unquote commercial, but it’s still in large part protected by the First Amendment, even though it has commercial aspects and is a for-profit business.

Say you wanted to incorporate Britney Spears’ voice without her consent in your podcast and then sold advertising on it but didn’t use her voice for the advertising. That’s still within bounds?

Probably. Maybe. But, you know, the way our legal system works is that we sort of invent the rules as we go along. So what’s going to happen is somebody’s going to take Britney Spears or Joe Biden or Donald Trump or whoever and use their voice in a podcast, and that person is going to sue, and then we’re going to see what the courts think of it. And it’s going to depend heavily on the specifics of the particular case.

With that said, if I were a podcaster and I was thinking about using an AI-generated voice, particularly of somebody famous, I would be very, very careful to disclose what I was doing. Before the rise of AI, if an actor was reading somebody’s letters or social media posts or whatever, you would have a disclosure saying this is a reenactment of such and such.

I would also be very careful in promoting the podcast to not make it seem as if Britney Spears is promoting the podcast. I don’t know why Britney Spears jumped into my head, but that’s where we are.

She can be our stand-in famous person.

She’s my stand-in famous person for today. I would be very, very careful about the advertising or the promotion of that podcast and not use anything more than is absolutely necessary. I would be very careful not to suggest in any way that Britney Spears actually appeared on the podcast, is associated with the podcast, is sponsoring the podcast, or endorses the podcast.

Apple Podcasts now requires in its new terms of service that you disclose whether you’re using AI in your podcast. I’m curious, since things get decided on a case-by-case basis, is this also something that may come up through the platforms, that they may decide these are the things you need to do in order to be in our space, and that’s how we see movement?

That’s certainly going to be part of it. I don’t think it’s going to be the whole story. There are going to be lawsuits, and those lawsuits are going to get decided, and those will give podcast creators, celebrities, and the rest of us some guidance as to what the rules of the road are.

With the platforms, there’s an ongoing dispute that we’re having sort of in society writ large about what are the roles of these platforms in policing the content that’s on them. We’re seeing that with X and the “anything goes” [attitude] versus some people pulling their blogs down off of Substack because they’re okay with Nazis. So I think there’s a wider discussion being had about what the role of the platforms is and what they’re going to regulate. I think that’s what you’re seeing with the terms of use from Apple and from YouTube.