US bill demands transparency over copyrighted works to train genAI

New bill introduced to Congress would require full disclosure from companies ahead of generative AI models being released | USPTO also publishes further guidance on AI use in IP proceedings before the office.

AI companies may have no choice but to disclose which copyrighted materials they used to create their generative AI models, if a “groundbreaking” new bill is passed in the US.

The bill was introduced to Congress this week (April 9) to require transparency from companies regarding their use of copyrighted work to train their generative AI models, such as ChatGPT.

Disclosures would need to be made at least 30 days prior to release—otherwise, companies would face fines.

US Democratic Representative Adam Schiff said that his Generative AI Copyright Disclosure Act would require AI companies to submit a notice to the Register of Copyrights prior to the release of a new generative AI system, with regard to all copyrighted works used in building or altering the training dataset for that system.

These requirements would also apply retroactively to previously released generative AI systems.

‘Crucial’ need for protections

Schiff, a former US federal prosecutor who led Donald Trump’s first impeachment trial said: “AI has the disruptive potential of changing our economy, our political system, and our day-to-day lives—but insisted that he is not against the use of AI per se.

“We must balance the immense potential of AI with the crucial need for ethical guidelines and protections,” he said.

The bill is “a pivotal step in this direction,” added Schiff. “It 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.”

According to the bill, submissions to the copyright register must contain “a sufficiently detailed summary of any copyrighted works used” either in the training dataset or to alter the training dataset, as well as that dataset’s URL for such dataset (in the case of a training dataset that is publicly available on the internet at the time the notice is submitted).

Industry backing

The proposed legislation is supported by multiple creative industry organisations, including the Screen Actors Guild-American Federation of Television and Radio Artists (SAG-AFTRA), the Recording Industry Association of America (RIAA), Professional Photographers of America (PPA), and the Directors Guild of America (DGA).

Ken Doroshow, chief legal officer of the RIAA, 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.”

Moiya McTier, senior advisor of the Human Artistry Campaign, described Schiff’s proposal as a “big step forward towards responsible AI that partners with artists and creators instead of exploiting them.

“AI companies should stop hiding the ball when they copy creative works into AI systems and embrace clear rules of the road for recordkeeping that create a level and transparent playing field for the development and licensing of genuinely innovative applications and tools.”

USPTO issues ‘reminder’ of AI rules

Meanwhile, the US Patent and Trademark Office (USPTO) today (April 11) published additional guidance in the Federal Register for practitioners and the public “to inform them of the important issues that patent and trademark professionals, innovators, and entrepreneurs must navigate while using AI in matters before the USPTO.”

The guidance, according to the office, “reminds individuals involved in proceedings before the USPTO of the pertinent rules and policies, helps inform those same individuals of the risks associated with the use of AI, and provides suggestions to mitigate those risks.”

Kathi Vidal, director of the USPTO, said that the guidance is “part of our work shaping AI policy, and encourages the safe and responsible use of AI to benefit the IP and innovation ecosystem."

“The requirements in existing USPTO rules serve to protect the integrity of our proceedings and to avoid delay and unnecessary cost, and they apply regardless of how a submission is generated.

“We will continue to listen to stakeholders on this policy and on all our measures to use AI responsibly and safely to democratise and scale US innovation, creativity, and entrepreneurship.”

AI-assisted patents

As obligated by President Biden’s Executive Order on Safe, Secure, and Trustworthy AI, the USPTO released guidance on AI-enabled innovations in February 2024, to explain how the office will determine whether such inventions can be patented.

Essentially, the guidance mandates AI-related patents to have “significant” human involvement, but Vidal emphasised that the guidance “embraces the use of AI in innovation and provides that AI-assisted inventions are not categorically unpatentable”.

The USPTO announced that it will also soon issue guidance on other aspects of patentability, and will issue recommendations for executive action related to copyright and AI.

In addition, the office said it is working with Secretary of Commerce Gina Raimondo and other federal agencies to guide US administration policy on AI and IP topics, such as the use of data in AI training; transparency and regulatory disclosures; trade secret protection; and the legal implications of AI-generated content.

The proposed legislation and the USPTO’s guidance are among broader attempts by multiple countries to balance AI innovation with protecting IP rights. These efforts are in part a response to the breakneck speed at which generative AI is developing, and the plethora of lawsuits being filed by authors, musicians, photographers and actors against the likes of OpenAI, Microsoft and Meta.

See WIPR’s recent webinar, AI and IP: What happens next?, featuring patent attorney Robert Jehan, Qualcomm’s Bob Giles, and Arun Hill from IP data firm Clarivate.

Did you enjoy reading this story?  Sign up to our free daily newsletters and get stories sent like this straight to your inbox