Generative AI

New Bill May Force AI Companies to Disclose Works Used to Train Models


Adam Schiff wants generative AI developers to unlock the black box.

The California Democrat announced Tuesday he had proposed a bill in the House of Representatives that would mandate generative AI developers to disclose all copyrighted works used in training their models. 

The bill, called the Generative AI Copyright Disclosure Act of 2024, states that companies or individuals that create or alter a training dataset used to create a generative AI system will need to inform the Register of Copyrights. When doing so, they will need to provide “a sufficiently detailed summary of any copyrighted works used in the training dataset…or to alter the training dataset.” 

To comply with the pending legislation, companies or individuals would need to file the notices at least 30 days prior to the release of the generative AI system. 

And pre-existing generative AI models won’t be exempt, either. The wording stipulates that companies that put generative AI systems on the market to consumers prior to the proposal of the bill would be required to file notices no later than 30 days after the effective date of the act, if it passes. 

Those who fail to comply would be slapped with a fine of at least $5,000. 

Schiff said the proposed legislation would help protect those working in creative industries. 

“AI has the disruptive potential of changing our economy, our political system, and our day-to-day lives. We must balance the immense potential of AI with the crucial need for ethical guidelines and protections,” Schiff said in a statement. “My Generative AI Copyright Disclosure Act is a pivotal step in this direction. 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.” 

Schiff’s proposal has received wide-spanning support from organizations and unions affiliated with people working in creative industries. Professional Photographers of America, the Recording Academy, the Directors Guild of America, the Copyright Clearance Center and others have put their stamp of approval on the bill. 

David Trust, CEO of Professional Photographers of America, said the organization’s 35,000 members would be excited to see the bill pass. 

“Photographers are particularly susceptible to having their works scraped by generative AI companies, as they must publicly exhibit their best work to attract clients. The urgency of this issue should not be understated as copyright holders suffer from the harsh reality of competing with their own works taken by generative AI companies to develop their systems,” Trust said in a statement. 

The bill could have far-reaching implications. For apparel and fashion brands, which copyright logos, images, designs and more, it could help stop potential infringement and clear up any murkiness around who—or what—has used copyrighted works for inspiration.  

The bill proposes that the Register of Copyrights would be required to issue regulations around the requirements within six months of the effective date, should the law pass. It would also have to create a “publicly available online database that contains each notice filed.” 

Schiff’s proposal comes at a time when generative AI creators like OpenAI, Microsoft and Stability AI face lawsuits accusing them of copyright infringement. While many of the cases have not yet been resolved, passage of this bill would reveal exactly what went into some of the most powerful generative AI models on the market today—at least relative to copyrighted works. 

It also comes on the heels of a newly minted partnership between the U.S. and the UK, focused on safe AI development. While the U.S. does not have an at-large, legally binding roadmap for AI in place as of yet, the EU recently adopted the landmark EU AI Act, which outlines requirements for technology developers and deployers doing business in its member countries.



Source

Related Articles

Back to top button