More than 10,000 professional actors, musicians, writers and other creators have signed a petition calling for AI to use their work without permission for training. British composer Ed Newton-Rex wrote the statement and set up the autograph collection. There are many well-known names among the signatories. They range from Hollywood stars like Kevin Bacon and Julianne Moore to record-selling musicians and composers like Radiohead’s Thom Yorke and Abba’s Björn Ulvaeus and best-selling authors Harlan Coben and Ted Chiang. The statement itself is short and to the point:
“The unlicensed use of creative works to train generative AI is a major, unjust threat to the livelihoods of the people behind these works, and should not be allowed.”
Essentially, the signatories fear that their copyrighted works will contribute without their permission to the vast amounts of data used to train generative AI models behind ChatGPT, Gemini, Meta AI and so many other text, image, audio creation tools and video. They claim it violates various intellectual property laws and regulations.
Along with individual celebrities, the statement has received the endorsement of many organizations in the creative space. SAG-AFTRA, the American Federation of Musicians, Universal Music Group and the International Publishers Association are just some of the people who have signed their support.
There is also the matter of compensation. Meta, for example, wrote hefty paychecks to celebrities for permission to use their voices with its new Meta AI assistant. Without that, these complaints cause problems, such as when OpenAI was accused of imitating Scarlett Johansson’s voice for ChatGPT following the movie Her.
Newton-Rex knows the AI space well, having previously worked on generative AI audio models at Stability AI. He has claimed that he left in part because he felt Stability AI went too far by relying on the fair use doctrine to train its models. He now runs Fairly Trained, which describes itself as a “nonprofit organization that certifies generative AI companies for fairer training data practices.”
Today we’re publishing a statement on AI training, already signed by more than 10,000 creators: “The unlicensed use of creative works to train generative AI is a major, unjust threat to the livelihoods of the people behind these works, and should not be allowed to be allowed .”Signatories include… pic.twitter.com/AqVaEThMs4October 22, 2024
AI pushback
This is far from the first such lawsuit. OpenAI alone has multiple pending cases from writers claiming ChatGPT infringes on their work, while Suno, Udio, and other AI music makers face lawsuits from the Recording Industry Association of America (RIAA) and major music labels. The complaints, legal and otherwise, are growing as the AI tools they rail against grow in popularity. They are one facet of the larger ethical and regulatory questions currently unanswered surrounding AI models and their training data.
You probably won’t see any impact on AI tools in the short term, but the signatories clearly hope to bring the debate about ethical AI training to the fore and shape the ultimate form of regulation and laws surrounding the practice. On its own it won’t amount to much, but combined with resolving legal challenges and new regulations it could be a factor in how AI companies design and build their models in the coming years, or whether the current system of compensation for creative work is much like today.