AI copyright issues take centre stage in George Carlin podcast dispute

‘Fair use’ may be used as defence in copyright suit against creators of ‘AI’ podcast Dudesy, but its success depends on wider legal battle being played out, say Kristen McCallion and Maggie LaPoint of Fish & Richardson 

At the start of the year, a comedy podcast called ‘Dudesy’ released an episode that purported to use artificial intelligence (AI) to impersonate the late comedian George Carlin. Carlin’s family swiftly denounced the show, and the Carlin estate filed a copyright infringement suit against the podcast and its creators. 

In the intervening weeks, questions have arisen as to whether AI was even involved in the creation of the show. However, these questions of fact pale in comparison to the increasingly complex questions of law that have been ushered in by the use of generative AI.

Dudesy debuts to litigious reviews

The Dudesy podcast purports to be written and directed by an AI of the same name. The show’s creators and hosts, Will Sasso and Chad Kultgen, have claimed that Dudesy, which was created by an undisclosed company, has access to all of their personal emails, texts, social media accounts, and online histories so that it can tailor the show “to their specific personalities”.

The podcast is no stranger to controversy. In April 2023, it released its first special, which depicted athlete Tom Brady delivering a stand-up comedy routine. Sasso and Kultgen claim that the AI Dudesy independently created the special without interference or prompting from them. After several weeks, counsel for Brady demanded that Sasso and Kultgen remove the video. Despite defending the special as protected parody, the creators complied with the demand.

Dudesy’s current legal troubles began with its second hour-long special titled ‘George Carlin: I’m Glad I’m Dead’. The show begins with a voice that says “Hello, my name is Dudesy, and I’m a comedy AI,” and then clarifies that the voice in the special is not George Carlin but is instead Dudesy’s impression of George Carlin that it “developed in the exact same way a human impressionist would”. 

Before launching into an eerily accurate impression of a ‘resurrected’ Carlin delivering a modern-day stand-up routine, the voice of Dudesy notes, “I listened to all of George Carlin’s material and did my best to imitate his voice, cadence, and attitude, as well as the subject matter I think would have interested him today.”

Shortly after the special’s release, the Carlin estate sued Dudesy and its creators, alleging copyright infringement and a violation of Carlin’s right of publicity. In an email to the New York Times, a spokeswoman for Sasso stated that Dudesy is not actually an AI but “a fictional podcast character created by … Will Sasso and Chad Kultgen”. Furthermore, she claimed that the special at issue was completely written by Kultgen.

In its complaint, the Carlin estate argued that the special constituted copyright infringement and an unlawful appropriation of Carlin’s identity even if the show creators “falsely ascribed the creation of the Dudesy special to the so-called ‘Dudesy AI’”. Counsel for the estate assured reporters that factual questions around how the special was created will be resolved through the discovery process. However, finding answers to the legal questions around generative AI may prove more difficult.

Generative AI: quantity and quality

At its most basic level, AI refers to the simulation of human intelligence in machines that are programmed to think and perform tasks in a way that mimics human cognitive functions, such as learning, problem-solving, decision-making, and perception. The recent headlines around AI generally refer to a subset of AI known as generative AI. 

Generative AI, the type of AI purported to be used by Dudesy, can create brand new content based on the data it was trained on. Unlike traditional algorithms that might simply categorise existing data, generative AI can “learn” from existing datasets and create new output, such as images, text, audio, and computer code.

The key to generative AI is data, and lots of it. AI is trained using massive amounts of data and algorithms that enable the system to learn from the input and generate output based on what it learned. Because the output is based entirely on the learnings from the input, the quality of the data used is as important as the quantity. Creators of AI programs must ensure that the material used to train their AI models is of the right type and quantity to produce the output they desire.

The right material for a wrongful use

However, the ‘right’ material may not be available for the taking. The legal battle between generative AI and the creators of the content on which it is trained will likely be fought primarily on two fronts: copyright infringement and the right of publicity.

As shown by the Dudesy case, generative AI may be trained using works that are protected by copyright. Carlin’s estate claims that Sasso and Kultgen unlawfully used George Carlin’s copyrighted works to train the Dudesy AI for the purpose of generating an imitation of Carlin’s stand-up comedy. Creating a dataset for training AI typically involves copying and distributing voluminous materials, which could constitute copyright infringement.

Some in the field believe that such uses should be considered fair use, which is a defence to copyright infringement. Indeed, statements that Kultgen wrote the special himself or that Dudesy AI “listened” to Carlin’s works and developed the special “in the exact same way a human impressionist would” are presumably to tee up a fair use defence. In its complaint, Carlin’s estate takes issue with this notion, alleging that the special “has no comedic or creative value”.

The right of publicity is also a flash point in the legal debate surrounding generative AI. Many generative AI systems deliver music, text, and image outputs in the ‘style of’ various artists, such as the AI-generated viral hit Heart on My Sleeve created in the style of Drake and The Weeknd. 

In this case, Dudsey’s special set out to mimic Carlin’s “voice, cadence, and attitude”. The Carlin estate alleges that Dudesy exploited Carlin’s identity by using his name and likeness in advertisements in the run-up to and immediately after the special’s launch, violating Carlin’s right of publicity under California statutory and common law.

A solution in sight?

While these debates rage, attempts are being made to fashion a fair compensation system. Shutterstock announced plans to create a fund to compensate individuals whose work it has sold to AI companies to train their models, while DeviantArt created a metadata tag for images shared on the web that warns AI researchers not to scrape their content. 

Others are considering a licensing system similar to that used in the music industry in which protected works may be used so long as intellectual property owners are fairly compensated. 

This issue is currently being litigated in several cases, and we will likely see many additional arguments and approaches before we settle on a solution as a legal community.

Kristen McCallion is a principal at Fish & Richardson. She can be contacted at: mccallion@fr.com.

Maggie LaPoint is an associate at Fish & Richardson. She can be contacted at lapoint@fr.com.