There will be no follow-up to that AI-generated George Carlin comedy special released by the podcast Dudesy. In January, Carlin’s estate filed a lawsuit against the podcast and its creators Will Sasso and Chad Kultgen, accusing them of violating the performer’s right to publicity and infringing on a copyright. Now, the two sides have reached a settlement agreement, which includes the permanent removal of the comedy special from Dudesy’s archive. Sasso and Kultgen have also agreed never to repost it on any platform and never to use Carlin’s image, voice or likeness without approval from the estate again, according to The New York Times.
The AI algorithm that Dudesy used for the special was trained on thousands of hours of Carlin’s routines that spanned decades of his career. It generated enough material for an hour-long special, but it did a pretty poor impression of the late comedian with basic punchlines and very little of what characterized Carlin’s humor. In a statement, Carlin’s daughter Kelly called it a “poorly-executed facsimile cobbled together by unscrupulous individuals.”
Josh Schiller, who represented the Carlin estate in court, told The Times that “[t]he world has begun to appreciate the power and potential dangers inherent in AI tools, which can mimic voices, generate fake photographs and alter video.” He added that it’s “not a problem that will go away by itself” and that it “must be confronted with swift, forceful action in the courts.” The companies making AI software “must also bear some measure of accountability,” the lawyer said.
This lawsuit is just one of the many filed by creatives against AI companies and the people that use the technology by training algorithms on someone’s work. Several non-fiction authors and novelists that include George R.R. Martin, John Grisham and Jodi Picoult sued OpenAI for using their work to train its large language models. The New York Times and a handful of other news organizations also sued the company for using their articles for training and for allegedly reproducing their content word-for-word without attribution.
Source link