A new Senate bill aims to make it easier for human creators to find out if their work was used without permission to train artificial intelligence, marking the latest effort to tackle the lack of transparency in generative AI development.
The Transparency and Responsibility for Artificial Intelligence Networks (TRAIN) Act would enable copyright holders to subpoena training records of generative AI models, if the holder can declare a “good faith belief” that their work was used to train the model.
The developers would only need to reveal the training material that is “sufficient to identify with certainty” whether the copyright holder’s works were used. Failing to comply would create a legal assumption — until proven otherwise — that the AI developer did indeed use the copyrighted work.
Sen. Peter Welch, D-Vt., who introduced the bill Thursday, said the country must “set a higher standard for transparency” as AI continues to integrate itself into Americans’ lives.
“This is simple: if your work is used to train A.I., there should be a way for you, the copyright holder, to determine that it’s been used by a training model, and you should get compensated if it was,” Welch said in a statement. “We need to give America’s musicians, artists, and creators a tool to find out when A.I. companies are using their work to train models without artists’ permission.”
The explosion of accessible generative AI technologies has triggered a slew of legal and ethical questions for artists, who fear these tools will enable others to recreate their work without consent, credit or compensation.
Though many major AI developers don’t publicly reveal their models’ training data, a viral Midjourney spreadsheet gave credence to artists’ concerns earlier this year when it listed thousands of people whose work was used to train its popular AI art generator.
Companies that rely on human creative labor have tried to take on AI developers as well.
In recent years, news outlets like The New York Times and The Wall Street Journal sued AI companies such as OpenAI and Perplexity AI for copyright infringement. And the world’s biggest record labels teamed up in June to take two prominent AI music-making companies to court, alleging that they trained their models on decades’ worth of copyrighted sound recordings without consent.
As legal tensions rise, more than 36,000 creative professionals — including Oscar-winning actor Julianne Moore, author James Patterson and Radiohead’s Thom Yorke — have signed an open letter urging the prohibition of using human art to train AI without permission.
No comprehensive federal legislation yet exists to regulate the development of AI, although several states have attempted to push through specific AI-related regulations, particularly around deepfakes. In September, California passed into law two bills aimed at protecting actors and other performers from unauthorized use of their digital likenesses.
Similar bills have been introduced in Congress, including the bipartisan “NO FAKES” Act, which would aim to protect human likenesses from nonconsensual digital replications, and the “AI CONSENT” Act, which would require online platforms to get informed consent before using consumers’ personal data to train AI. Neither has gotten a vote so far.
In a news release, Welch said the TRAIN Act has been endorsed by several organizations — including the Screen Actors Guild-American Federation of Television and Radio Artists (SAG-AFTRA), the American Federation of Musicians, and the Recording Academy — as well as major music labels — including Universal Music Group, Warner Music Group and Sony Music Group.
Only a few weeks remain in this Congress, however, and members are focused on must-pass priorities like avoiding a government shutdown on Dec. 20. Welch’s office said he plans to reintroduce the bill next year, as any unpassed legislation will need to be reintroduced in the new Congress when it convenes in early January.
