YouTube says fans want to put themselves in AI videos. For music, the next step is putting artists in places they never agreed to be

YouTube's Culture & Trends team published a report on generative video and personal creativity on 20 October. Its headline is that ordinary people, not just studios and spammers, are now using AI to make videos about themselves. Music Ally picked it up on 23 October and pointed to where the trend is already showing up in music: fans dropping one star into another star's music video. Here is what the report says, what it leaves out, and what it means for artists.

What does the report say?

The report is called "Human Creativity and AI". YouTube says the team reviewed the year's major generative video trends and ran surveys with the research firm NRG. The figures it quotes come from one survey, footnoted as "Google/ NRG, YouTube Trends Survey, U.S. April 2026, N=694 online people aged 14–44". The blog post also says "thousands of people" were asked about their views on generative AI, but the numbers it publishes trace back to that sample of 694.

The key numbers:

• 72% of the surveyed video posters say they use AI on their videos or images.
• 61% of people who posted video in the past year say they are interested in using AI to insert themselves into content.
• More than two-thirds of people who post AI videos say they have featured themselves in them.

YouTube's framing is that, "amid skepticism about the benefits of generative video, many are beginning to use it in a personal capacity". Hobbyists, it says, are "removing the constraints of reality" to explore their passions. YouTube describes it as "digital costume changes", according to Music Ally: people imagining themselves in other roles.

Treat the numbers with some care. This is a US-only online sample of under 700 people, commissioned by the platform that benefits if creators make more AI video. It shows a direction, not the size of a market.

Where does music come in?

YouTube's blog post summarising the report doesn't mention music. Music Ally's example does. A fan-made Instagram Reel put Ariana Grande into LISA's "SaWaDiKa" video. According to Music Ally, it drew 82.2k likes and 24.7k shares, more than 100 times the engagement of the creator's usual posts. Some accounts, it says, now focus on this kind of artist-swap content, and it notes that such videos can carry a song to another artist's fanbase.

Music Ally then suggests three things artist teams can do:

• Know the tools. The Grande and LISA video was made with Higgsfield, which Music Ally reported had more than 15 million users back in January. It suggests teams try the tool on their own artist's assets.
• Invite fans in. It cites Ghost's "Satanizer", which let fans face-swap themselves into the band's "Satanized" video.
• Brief AI creators. Teams could ask fans to make AI content around a release, or commission AI creators to build on a music video.

What protection do artists have on YouTube?

Some, but it is narrow. YouTube's likeness detection scans new uploads for faces that match an enrolled person and may be AI-altered, and lets that person ask for removal through the privacy complaint process. In April YouTube opened it to talent agencies, management companies and their clients, including people without a YouTube channel. The help page says it works on faces for now, with voice matching still rolling out to some users. We walked through setting it up, and its limits, in our likeness protection guide.

That tool only covers YouTube. The Grande and LISA example was an Instagram Reel. A removal request also depends on someone being enrolled, noticing, and choosing to complain, and YouTube says it weighs parody and satire before removing anything. As we found when Sony asked for more than 260,000 AI tracks to be taken down, the artists with the most to lose from this kind of content are often the ones with the least capacity to police it.

What we think

The report, as YouTube summarises it, measures demand and skips the rights question. It counts people who want to put themselves into AI video. The music version of the trend is people putting other people, real artists, into video those artists never made. YouTube's blog post summarising the report doesn't use the words "consent", "likeness" or "label". For a platform that also sells likeness detection to the entertainment industry, that silence is worth noticing.

Self-insertion and artist-swapping need different rules. A fan putting their own face into a music video is closer to a filter or a dance challenge: the artist's work is the backdrop and the fan is the subject. An artist-swap video makes a real performer appear to perform something they didn't. One is participation, the other is impersonation, even when it is affectionate. Music Ally's Ghost example works precisely because the band built the tool and chose the terms. We think that is the model to copy: official, opt-in face-swap tools that let fans put themselves in, rather than encouraging open season on any artist's face.

Our non-obvious take: the second artist pays for the first artist's promotion. The Grande and LISA Reel is a good example. It plausibly helped "SaWaDiKa" reach Grande's fans. But it is Grande's likeness doing the work, and she gets no credit, no payment and no say. If labels start briefing AI creators to make swap content, as Music Ally floats, they are asking creators to borrow other artists' faces for marketing. A label that would sue over an unlicensed sample should not commission an unlicensed face. Teams that want this kind of reach should restrict it to their own artist's assets and fans' own faces, and say so in the brief.

What to do now. If you are an artist or manager: enrol in YouTube's likeness detection if you are eligible, keep an eye on Reels and TikTok by hand, and decide in advance what fan AI content you are happy with, so you can say so publicly. Places like Denmark are already moving toward making the poster prove consent. A clear public line from the artist now makes any later takedown request easier to argue.