Does Your Talent Release Address AI?

Talent releases have always addressed what a company can do with someone’s image, likeness, or voice. AI has made that question a lot more complicated.

That came up recently when my son was doing a shoot for a company I genuinely admire. The photographer is one of my closest friends—and the mom of his best friend. When she sent me the talent release, I looked for an AI clause.

There wasn’t one.

I didn’t think anyone involved in the shoot had any intention of using my son’s image or voice in a way I would be uncomfortable with. We were doing this as a favor for a friend, and all of the talent were young kids.

And still. All of the talent were young kids.

So my mom-and-lawyer brain went to the same place: Could my son’s image or voice ever be used to create something I never agreed to?

If I were representing the company (and I often do), I’d be asking exactly the same question.

AI training is only part of the issue.

I raised the AI problem with the photographer. She immediately understood the concern and took it back to the company.

A few days later, they proposed language prohibiting the use of the kids’ images to train AI. It was a good start. But it didn’t address AI-generated or manipulated content—the thing I was actually worried about.

Those are two different things.

You can stop someone from using content to train, fine-tune, or otherwise develop an AI system without saying anything about whether AI can be used to create something new from that content.

For me, that was the bigger concern. Could AI be used to change my son’s appearance? His voice? Make it look or sound like he did or said something he never actually did or said?

So I proposed language that addressed both.

My friend later told me the process had been highly educational for her and that she planned to use the AI language in her future contracts.

“No AI” wasn’t the answer either.

I could have proposed a blanket prohibition on using AI in any way involving my son’s image, likeness, or voice. I didn’t.

Companies need to be able to do ordinary editing. AI is also increasingly built into tools people use every day, sometimes without even thinking about it as AI. And if a company posts content on its website or social media, it can’t control everything that happens to that content afterward.

So the language I proposed focused on the thing I actually cared about: the company knowingly using AI or other generative technology to create or materially alter my son’s likeness, including his voice or other identifying characteristics, in a way that materially changed his appearance, voice, actions, or identity.

At the same time, I wanted the language to protect the company so ordinary editing and uses outside its control wouldn’t create a problem.

That was better for both sides.

Social media makes this even more complicated.

Social media is a huge part of most modern brands’ marketing strategy. Once a brand puts photos or videos on a public platform, other people may be able to copy them, scrape them, manipulate them, or use them in ways neither the company nor the person in the content ever intended.

That’s a problem for the person in the content. But it can also be a problem for the brand. A company probably doesn’t want someone taking content from one of its campaigns and using AI to turn it into something the company never created or approved either.

A talent release can say what the company is allowed to do with someone’s image, likeness, or voice. But the company doesn’t control what other people—or bots—do with that content on the internet.

That’s why the language I proposed focused on what the company knowingly does. It protects the talent without making the company responsible for every bot on the internet.

If you don’t understand how the AI clause works, ask someone.

I always recommend that creators and talent (and their parents) read the release carefully and make sure they understand how it works before they sign anything.

Some brands may actually want the right to use AI to repurpose talent content. Maybe they want to alter an image, generate new content from existing footage, change a voice, or create something the talent never actually recorded.

If you’re the talent—or the parent of a child who is—you should know whether you are being asked to agree to that.

Does the language prohibit AI training but allow generative uses? Does it give the company the right to create new content using your likeness or voice? Does it require additional consent? What happens to those rights after the original campaign is over?

The question isn’t whether the release has an AI clause. It’s what that AI clause says the company can and cannot do with your image.

In this story, the original release didn’t address AI. Now it does. And both sides are better protected because of it.