Representative Adam Schiff (D-Calif.) introduced new legislation in the US House of Representatives on Tuesday (April 9) that, if passed, would require artificial intelligence companies to disclose which copyrighted works were used to train the models them, otherwise they will face a financial penalty. Called the Generative AI Copyright Disclosure Act, the new bill would apply both to new models and retroactively to previously released and used AI generation systems.
The bill requires that the full list of copyrighted works on an AI model's training dataset be submitted to the Copyright Office no later than 30 days before the model becomes available to consumers. This will also be required when the training data set for an existing model is modified in a significant way. Financial penalties for non-compliance will be determined on a case-by-case basis by the Copyright Office, based on factors such as the company's history of non-compliance and the size of the company.
Productive AI models are trained on up to trillions of existing projects. In some cases, datasets, which can include anything from movie scripts to news articles to music, are licensed by copyright holders, but often these models will “sweep” the internet for large chunks of content, some of which are protected by copyright, without the consent or knowledge of the author. Many of the world's largest AI companies have publicly defended the practice, calling it “fair use,” but many in the creative industries argue that it is a form of widespread copyright infringement.
The debate has sparked several lawsuits between copyright holders and AI companies. In October, Universal Music Group, ABKCO, Concord Music Group and other music publishers filed a lawsuit against artificial intelligence giant Anthropic for “illegal exploitation” of their copyrighted song lyrics to train AI models.
“In the process of building and operating artificial intelligence models, Anthropic illegally copies and distributes vast amounts of copyrighted works,” lawyers for the music companies wrote at the time. “Publishers are embracing innovation and recognizing the great promise of AI when used ethically and responsibly. But Anthropic violates these principles on a systematic and widespread basis.”
While many in the music industry are also calling for compensation and the ability to opt-in or opt-out of a data set, this bill focuses only on requiring transparency with copyrighted training data. However, it has garnered support from several music industry groups, including the Recorded Industry Association of America (RIAA), National Music Publishers' Association (NMPA), ASCAP, Black Music Action Coalition (BMAC), and the Human Artistry Campaign.
It is also supported by other creative industry groups, including Professional Photographers of America, SAG-AFTRA, Writers Guild of America, International Alliance of Theatrical Stage Employees (IATSE) and others.
“Artificial intelligence has the potential to change our economy, our political system and our daily lives,” said Representative Schiff in a statement. “We must balance the enormous potential of artificial intelligence with the critical need for ethical guidance and protections. My Generative AI Copyright Disclosure Act is a critical step in this direction. It supports innovation by protecting the rights and contributions of creators, ensuring they know when their work contributes to AI training datasets. It's about respecting creativity in the age of artificial intelligence and marrying technological progress with fairness.”
Some rights groups also opposed the bill's introduction.
“Any effective AI regulatory regime must begin with one of the most fundamental building blocks for effective copyright enforcement — comprehensive and transparent recordkeeping,” adds the RIAA's chief legal officer Ken Doroshow. “The RIAA applauds Congressman Schiff for leading on this urgent and fundamental issue.”
“We commend Congressman Schiff for his leadership on the Genetic Artificial Intelligence Copyright Disclosure Act,” NMPA President/CEO David the Israelite he said. “AI only works because it mines the work of millions of creators every day, and it's important that AI companies disclose exactly which projects train their data. This is a critical first step towards ensuring that AI companies are fully licensed and that songwriters are fully compensated for the work used to power these platforms.”
“Without transparency about using copyrighted works to train AI, creators will never be fairly compensated and AI tech companies will continue to steal from songwriters,” ASCAP CEO Elizabeth Matthews he said. “This bill is an important step toward ensuring that the law puts people first, and we thank Congressman Schiff for his leadership.”
“Protecting the work of music creators is essential, and it all starts with transparency and monitoring of the use of copyrighted material in genetic artificial intelligence,” co-chair of the Black Music Action Coalition (BMAC). Willie “Prophet” Stigers he said. “BMAC hopes that the Representative of Vice President Schiff's AI Copyright Disclosure Act will help rally support for this mission and that creator and creator rights continue to be protected and preserved.”
“Congressman Schiff's proposal is a big step forward toward responsible AI that works with artists and creators rather than exploiting them,” Senior Advisor to the Human Artistry Campaign Dr. Moiya McTier he said. “AI companies should stop hiding the ball when copying creative work on AI systems and adopt clear record-keeping rules that create a level and transparent playing field for developing and licensing truly innovative applications and tools.”
from our partners at https://www.billboard.com/business/legal/federal-bill-ai-training-require-disclosure-songs-used-1235651089/