He went on to talk about his playing career, defecating, World War 3, and much more. But here’s the thing: Tom Brady never said these things.  

Tom Brady’s “comedy routine” is the product of Will Sasso and Chad Kultgen, the hosts of the podcast Dudesy, who used generative artificial intelligence to create a one-hour-long comedy routine that mimics the voice of NFL great Tom Brady.

This comedy routine was available as a perk on Patreon to Dudesy subscribers, but the comedy bit quickly proliferated on the internet. Other podcasts, including the Pat McAfee Show, commented on how life-like and realistic Tom Brady’s comedy routine was.

Unfortunately, this came at a cost; TMZ reported that Sasso and Kultgen received a cease-and-desist letter from Brady’s attorneys, alleging that Dudesy infringed on Tom Brady’s personality rights, namely the use of his voice, to create this comedy special.

According to Sasso and Kultgen, the same cease-and-desist letter threatened a personality rights infringement lawsuit and demanded that the fake comedy special be removed from all social media platforms and anywhere the comedy routine was disseminated. Sasso and Kultgen complied.

The Broader IP

In April 2023, the German tabloid magazine Die Aktuelle published an “interview” with famed F1 racer Michael Schumacher. According to the tabloid, this was the first time Michael Schumacher publicly spoke to the media since suffering a near-fatal brain injury in December 2013 while skiing. In the article with Die Aktuelle, Schumacher commented on his skiing accident that led him to step away from F1 racing. However, at the conclusion of the article, the tabloid magazine disclosed that this interview with Michael Schumacher was, in fact, completely AI-generated. Michael Schumacher’s family was extremely upset, and rightfully so, and planned to pursue legal action against the publication. The editor in chief of the magazine was fired.

Generative AI platforms like ChatGPT, Midjourney, and Dall-E have proliferated at an exponential rate. As of January 2023, OpenAI reported nearly 100 million users signed up for ChatGPT since its public launch in November 2022. Dall-E reached 1 million users after 2.5 months. According to Statista, the adoption rate of generative AI in the United States in the workplace is averaging 28% across Gen X, Millennials, and Gen Z. There’s also no denying how innovative generative AI technology truly is. You can go into Chat GPT and ask the system to write a poem about what a great basketball player Michael Jordan was and the AI will do all the work for you. AI systems can generate artwork of things we could never conceive of in real life. But while the creative potential of generative AI is clear, there are also plenty of risks and unanswered questions. There is a great deal of legal ambiguity that stems from AI creation and use.

This isn’t the first time though that generative AI has gotten into hot water with an athlete’s personality rights.”

and Constitutional Implications

Most legal issues involving generative AI stem from copyright law. In a recent decision, the US Copyright Office made it clear that AI-generated works, including AI-generated artwork, will not receive any copyright protection because AI-generated works lack the human authorship required for protection under copyright. But perhaps one of generative AI’s biggest foes is the personality rights law.  

Unlike copyright, trademark, patent, trade dress, and trade secret law, personality rights are the only area of intellectual property law that is not regulated at the federal level.  Currently, there are only 25 states that have adopted personality rights legislation, also known as the “right of publicity.”.  Other states adopt a common law standard.  Regardless, the standard is the same across the board: personality rights involve the use of one’s name, image, and likeness for commercial use. If anyone uses someone’s personality rights for commercial use without that person’s consent, that qualifies as a personality right infringement. One of the cornerstone personality rights infringement cases is the iconic 9^th Circuit decision of Midler v. Ford Motor Co.  

In this case, Ford approached Bette Midler and her agent for permission to use her song “Do You Want to Dance” in a commercial campaign for Lincoln Mercury cars.  Bette Midler denied permission so instead, Ford hired Midler’s backup singer Ula Hedwig to sing the song like Bette Midler. The commercial went live and Midler sued Ford, arguing that recreating her voice was a direct infringement of her personality rights under the California Celebrities Act. The 9^th Circuit sided with Midler, arguing that use of her voice (albeit a recreated version) to solicit car sales was an infringement of her personality rights. The 9^th Circuit further added that Midler’s voice is as much a part of her identity as her actual name, image, and likeness. 

of Generative AI

Bringing this back to generative AI, it’s evident that using AI to recreate one’s personality rights, including one’s voice, can create serious legal issues. Cases like Midler v. Ford serve as a warning signal that using AI to imitate a celebrity or athlete is an infringement-worthy activity if you are using that material for commercial purposes, like an advertisement. And then you have situations like the Brady-AI comedy routine, where Tom Brady’s voice was recreated to make crude jokes. While some legal critics may argue that such activity infringes on Brady’s personality rights, a majority, including myself, may argue that such activity is freedom of expression that is protected under the First Amendment. Parodying a public figure is something that the US Supreme Court has regarded as a fully protectable interest under the Constitution. Essentially, Dudesy’s Tom Brady AI-generated comedy routine is no different than SNL doing a parody of Donald Trump or Joe Biden if the whole point is to parody their personas. That type of commentary is an entirely protectable interest under the First Amendment.

But what about Michael Schumacher’s fake interview with a German tabloid magazine? I would lean towards saying that, in the United States, such behavior is defamatory. Defamation is a cause of action that falls under the First Amendment of the US Constitution and it occurs when someone makes a false statement about someone else and damages their reputation. For public figures, they have the burden of proving actual malice, meaning that the offender knew that the statement was false and with reckless disregard still made the defamatory statement. While the Schumacher interview would be governed under German law, it would be hard to dismiss that such behavior here in the United States meets the threshold of defamation. Imagine if a newspaper or news outlet framed an AI-generated interview with an athlete as a real one that paints said athlete in a bad light. Not only is there an ethical dilemma, but such behavior triggers a cause of action of defamation.

It’s evident that using AI to recreate one’s personality rights, including one’s voice, can create serious legal issues.”

Use Among Athletes

On the contrary, I think the innovation of generative AI opens the door wide open to sparking creativity and ideas. As it pertains to the sports industry, generative AI presents a plethora of great opportunities. However, there are more questions than answers concerning artificial intelligence and its place in the legal field. The lack of regulation on the local and federal levels also doesn’t help the matter. At the end of the day, we must regard the AI landscape as the Wild West. It is a poorly regulated terrain, but I think prior case law and any impression of existing statutory language give us a roadmap of how to approach regulating this budding industry. In the meantime, athletes should be cautious of working with any brands/entities using generative AI and should ensure that they are legally protected by way of a formal contract.