US bill would require AI companies to disclose copyrighted work in training models

Growing prevalence of AI in creative space makes legislative guidance on use of copyrighted work to train software ‘seem inevitable’
Portland, OR, USA - Dec 18, 2022: Webpage of ChatGPT, a prototype AI chatbot, is seen on the website of OpenAI, on a smartphone. Examples, capabilities, and limitations are shown before a new chat.

Tada Images; Shutterstock

US legislation that would require AI companies to disclose all of the copyrighted work they used to train their generative AI models has been introduced to the US House of Representatives.

The Generative AI Copyright Disclosure Act, introduced by California democratic congressman Adam Schiiff, would require if passed, a notice to be submitted to the Copyright Office prior to the release of a new generative AI system, listing out all copyrighted works used in building or altering the training dataset for that system.

The bill’s requirements would also apply retroactively to previously created generator AI systems. The notice would be filed no later than 30 days before the generative AI system was made public or face a financial penalty of not less than $5,000.

Generative AI “learns” and is trained on documents and artefacts that already exist online and allows users to input a variety of prompts to generate new content such as text, images, videos and other media. 

AI companies like ChatGPT creator OpenAI, have been subject to a growing number of lawsuits from novelists and performers like Sarah Silverman and US media organisation New York Times who have accused them of using their copyrightable material without their permission to train their software.

Schiff said the bill “champions innovation” while safeguarding the rights and contributions of creators, ensuring they are “aware when their work contributes to AI training datasets. This is about respecting creativity in the age of AI and marrying technological progress with fairness”.

Schiff’s bill has generated significant support from the creative community.

Ken Doroshow, chief legal officer from the Recording Industry Association of America, said any effective regulatory regime for AI must start with one of the “most fundamental building blocks of effective enforcement of creators’ rights – comprehensive and transparent recordkeeping” and applauded Schiff for “leading on this urgent and foundational issue”.

Meredith Stiehm, president, Writers Guild of America West, said that the bill was an “important first step in addressing the unprecedented and unauthorised use of copyrighted materials to train generative AI systems. Greater transparency and guardrails around AI are necessary to protect writers and other creators”.

A slew of other US organisations and unions from the creative industries have come out in support of the bill including Directors Guild of America, Authors Guild, National Association of Voice Actors, Concept Art Association, Professional Photographers of America, Screen Actors Guild-American Federation of Television and Radio Artists and Writers Guild of America.

Jennifer A Mauri, a senior associate at Los Angeles-headquartered Michelman & Robinson, pointed out that the bill is still in its “absolute infancy”. It will need to be approved by one of the House of Representatives standing committees, and if it is, then will need to be debated and voted on by the House. If it passes, she said, it is then sent to the US Senate where it essentially goes through the same process. If approved by the Senate, it then goes to the President for his signature or veto.

She added that a majority of bills that are introduced never make it all the way through the process to become law and, while it remains to be seen whether this particular bill will be enacted, “given the growing prevalence of AI in the creative space it seems inevitable that the legislature will step in to provide guidance so that people on both sides of this issue can determine the boundary of what is and isn’t protected under copyright law”.

Email your news and story ideas to: [email protected]