During a segment on The Daily Show on Wednesday night, guest host Sarah Silverman dug further into her decision to sue the OpenAI language model ChatGPT for copyright infringement.
Back in July, Silverman joined a class action suit alongside authors Richard Kadrey and Christopher Golden, claiming that ChatGPT had stolen content from her book The Bedwetter without permission.
“While AI has helped open many avenues for many new works, there’s just one problem: they’re not new works,” she told the Daily Show audience. “What these programs do is scrape text and images from existing works and feed it into their system to create copycats.”
ADVERTISEMENT
The comedian demonstrated how AI can warp existing works of art by showing a side-by-side comparison of the Mona Lisa with an artist’s rendering of how the famous mystery woman would look today. (Spoiler alert: She would apparently have huge boobs.)
“These programs are printing money. ChatGPT is on track to make a billion dollars just this year alone, which is great for them, but the problem is that these companies are using artists’ work without consent or credit or payment,” Silverman continued. “And I’ve had firsthand experience with this theft, because one of the 100,000 books used to train ChatGPT was my book, The Bedwetter, available wherever books are sold.”
One of the lawsuits Silverman joined this summer claims that when ChatGPT is asked to summarize copyrighted material such as her book, a “derivative” version of that work is produced. The suits further suggest that the mere existence of these AI models are illegal under the Copyright Act since they need to be fed with potentially copyrighted information in order to work as expected.
“By the way, that book is about my actual life,” Silverman added on The Daily Show. “My jokes, my life, my experiences, my pain. ChatGPT stole all of it—and didn’t even have the courtesy to give me Mona Lisa tits.”
