Subscriber Benefit
As a subscriber you can listen to articles at work, in the car, or while you work out. Subscribe NowMore than 10,500 creative professionals, including Thom Yorke from Radiohead, actress Julianne Moore, and Nobel Prize-winning novelist Kazuo Ishiguro, have signed an open letter condemning “unlicensed use of creative works” to develop artificial intelligence systems such as OpenAI’s ChatGPT.
Use of creative work without a license for AI development is “a major, unjust threat to the livelihoods of the people behind those works, and must not be permitted,” the brief, 29-word letter says.
OpenAI and other tech companies need text, images, video and other material to train the algorithms that power chatbots and other AI systems. That data has often been scraped from the internet without consent, compensation or credit.
Tech companies have argued this practice is protected as “fair use” under copyright law, but content owners and publishers have increasingly fought back. They have asserted in lawsuits and pleas to regulators that AI developers using their work have illegally infringed on their copyright protections.
“This question of creators’ rights is incredibly pressing,” said Ed Newton-Rex, a former AI executive and music composer who helped organize the letter released Tuesday and is now CEO of the nonprofit Fairly Trained, which certifies tech companies for data practices that support creators’ rights. “Right now, it’s important to send a message.”
Several high-profile lawsuits against AI companies over data use are working through the courts. Regulators in the United States and Britain are debating whether copyright exemptions should be created for artificial intelligence projects, including the possibility of allowing AI companies to scrape data unless artists and publishers opt out.
“There’s a potential outcome where it’s essentially too late to do anything about it,” said Newton-Rex, who resigned from his role at high-profile start-up Stability AI last year because he disagreed with its position that training algorithms on copyrighted work constituted fair use.
The letter released Tuesday was signed by actors including Kevin Bacon and Melissa Joan Hart; comedians Kate McKinnon and Rosie O’Donnell; authors including Ann Patchett and Emily St. John Mandel; songwriter Billy Bragg and ABBA’s Björn Ulvaeus; and a slew of award-winning composers who worked on movies such as “Avatar” and “The Life of Pi.”
Industry figures including the head of trade organization Songwriters of North America and the CEO of Hachette Book Group also signed the letter.
Fairly Trained’s website lists music label Universal Music Group, trade group Association of American Publishers, and creative trade unions SAG-AFTRA and the Authors Guild on its website as supporters of its work, but they have not provided financial backing, Newton-Rex said. The nonprofit supports itself using fees it charges companies to certify their data use for AI projects as fair to content creators.
The letter comes after industry-specific protests against unauthorized use of creative work for AI systems. Those have included musical artists such as Billie Eilish and Nicki Minaj, Hollywood documentarians, and a group of more than 15,000 authors including Margaret Atwood and Jonathan Franzen.
At the same time, U.S. courts are allowing claims in a number of lawsuits against AI companies over data use to proceed.
In July, a U.S. district judge in California allowed a claim of copyright infringement to move forward in a lawsuit filed against OpenAI by writers including comedian Sarah Silverman. An allegation of unfair business practices in the suit was dismissed. In August a different U.S. district judge in the state allowed key parts of a lawsuit against image generation start-ups Stability AI and Midjourney to advance. Experts warn it will be challenging for plaintiffs to win such cases because they may have to prove that the outputs from AI systems are imitating specific works, not just using them as one input among many that influence the system’s behavior.
Some content publishers have taken a different tack, signing agreements with AI developers to provide access to their data in return for payment or other benefits.
Please enable JavaScript to view this content.
All beginning to sound like contrived chaos.