George Carlin poses while promoting his book “All My Stuff” in 2007.
Mark Maines/Getty Images
A settlement has been reached between the estate of George Carlin and the podcast makers who used generative artificial intelligence to impersonate the late comedian and his style in an unauthorized special.
Will Sasso and Chad Koltgen, hosts of the podcast dudeGeorge Carlin's estate notified the court on Tuesday of an agreement to resolve the case. Under the deal, an injunction will be issued preventing further use of the video, which has already been removed and was filmed in violation of the comic's rights, says Josh Schiller, an attorney for the estate. Additional terms of the agreement were not disclosed. Schiller declined to comment on whether there were financial damages.
This settlement represents what is believed to be the first resolution of a lawsuit related to the misappropriation of a celebrity's voice or image using artificial intelligence tools. It comes as Hollywood is sounding the alarm about technology being used to exploit the personal brands of actors, musicians and comedians, among others, without consent or compensation.
“This sends a message that you have to be very careful about how you use AI technology, and respect people's hard work and good intentions,” Schiller says. He adds that the deal “will serve as a blueprint for resolving similar disputes moving forward when AI technology violates the rights of an artist or public figure.”
“This case serves as a warning about the risks posed by artificial intelligence technologies and the need for proper safeguards not only for artists and creators, but for every human being on Earth,” author and producer Kelly Carlin, daughter of George Carlin, said in a statement.
The legal battle stems from an hour-long special titled George Carlin: I'm glad I'm dead, which was released in January on the YouTube podcast channel. In the episode, AI-created George Carlin, who imitates the comedian's signature style and cadence, narrates commentary on AI-generated images and addresses current topics such as the rise of reality television, streaming services, and AI itself.
Described as a “first-of-its-kind media experience,” the show revolves around the use of artificial intelligence software called “Dudesy AI” — which has access to most of the host’s personal records, including text messages and social media. Accounts and browsing history – for writing episodes in the style of Sasso and Kultgen.
Schiller says the podcasters contacted George Carlin's estate and offered to remove the video and agree not to republish it on any platform moving forward. He adds: “We wanted to move on from this quickly and honorably.” [Carlin’s] The legacy and reclaim it by getting rid of this.”
The lawsuit alleged copyright infringement due to unauthorized use of the comedian's copyrighted works.
At the beginning of the video, it's explained that the AI software that created the specials George Carlin absorbed over five decades, owned by the comedian's estate, as training material.
The complaint also alleged violations of right of publicity laws for use of George Carlin's name and likeness. He referred to the promotion of the special as a George Carlin AI-generated sequel, in which the deceased comedian was “resurrected” using AI tools.
This wasn't the first time Dudesy had used artificial intelligence to impersonate a celebrity. Last year, Sasso and Kultgen released an episode featuring an AI-generated Tom Brady performing a comedy routine. It was removed after the duo received a cease and desist letter.
In the absence of federal laws covering the use of artificial intelligence to mimic a person's appearance or voice, a host of state laws have filled the void. However, there is little recourse for those living in states that have not passed such protections, which has prompted Hollywood to apply pressure.
That prompted a bipartisan coalition in the House of Representatives to introduce a long-awaited bill in January that would ban the publication and distribution of unauthorized digital copies, including deepfakes and audio cloning. The legislation aims to give individuals the exclusive right to consent to the use of their image, voice and visual likeness by granting intellectual property rights at the federal level. Under the bill, unauthorized uses would be subject to severe penalties and lawsuits would be possible by any person or group whose exclusive rights were affected.
In March, Tennessee became the first state to pass legislation specifically aimed at protecting musicians from unauthorized use of artificial intelligence to imitate their voices without permission. The ELVIS Act builds on the state's longstanding right of publicity law by adding an individual's “voice” to the sphere it protects. California has not yet updated its statute.
Tuesday's deal arrives on the heels of OpenAI preparing to launch a new tool that can recreate a person's voice from a 15-second recording. When given a recording and a text, he can read that text with a voice from the recording. The Sam Altman-led group said it is not releasing the technology to better understand potential harms, such as using it to spread misinformation and impersonate people to facilitate fraud.
Amid the rise of AI voice simulators, there is debate about whether platforms that host infringing content should be subject to liability. Under the DMCA, platforms like YouTube can take advantage of certain safe harbor provisions as long as they take certain steps to remove such potentially infringing content. Artist advocacy groups called for a review of the law.
“This is not a problem that will go away on its own,” Schiller said in a statement. “It must be met with swift and forceful action in the courts, and AI software companies whose technology is weaponized must also bear some degree of accountability.”
