The bill promises transparency but could threaten innovation.
The Generative AI Copyright Disclosure Act, championed by Rep. Adam Schiff, is all about bringing transparency to the world of generative AI but does it leave the music industry and tech companies grappling with a complex new reality?
Here’s the good and the ugly on how it could effect us:
- Companies need to notify the Register of Copyrights before launching a new generative AI system.
- They must disclose all copyrighted works used in training their AI.
- It applies retroactively, meaning even past AI systems must be transparent about their training data.
- No clarity on how AI companies should deal with complicated attribution and copyright issues.
- Mysteries about whether you need permission from just labels… or if you need consent from artists, songwriters, and publishers as well.
Rep. Schiff laid out:
“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.”
Ai companies respond with: “uh… how?”
Recording Industry Gives a Thumbs Up
This bill has struck a chord with the creative community. Here’s what some key players had to say:
- Ken Doroshow, Chief Legal Officer at the RIAA, praised the bill for its push for comprehensive record keeping, essential for protecting creators’ rights.
- Elizabeth Matthews, CEO of ASCAP: "Without transparency around the use of copyrighted works in training artificial intelligence, creators will never be fairly compensated and AI tech companies will continue stealing from songwriters. "
- David Israelite, President/CEO of NMPA: "AI only works because it mines the work of millions of creators every day… "
- Mike O’Neill, President & CEO of BMI: "The Generative AI Copyright Disclosure Act would help ensure transparency and accountability when AI systems exploit the work of creators’ without their knowledge or consent."
But It’s Not All Sunshine and Rainbows
While the creative community is on board, the bill has its critics:
- Logistical Nightmares: Eric Goldman, a tech law professor, called the bill impractical, pointing out the challenges in tracking and disclosing every copyrighted work used in AI training.
"I don’t even really understand what he’s trying to do. Trying to disclose your sources when building an AI generative index—it’s not possible”
- Innovation Hurdles: Some argue that the bill could slow down AI development, as developers get bogged down in the paperwork of tracking millions of copyrighted pieces.
Bills Bills and More Bills…Why It Matters For Sound Ethics
The US AI copyright act, EU AI Act, UK Artificial Intelligence (Regulation) Bill, and others focus on implementing transparency and permissions into training data for generative AI companies. They all have similar critics and negative effects of slowing down innovation to deal with copyright clearance and regulation.
Sound Ethics is committed to finding solutions for these unique problems by:
- Making available ethical training data sets with attribution for licensing
- Transparent training reports that balance proper attribution while protecting valuable IP
- Developing standards on metadata attribution and certifying the source and history (or provenance) of music data
- Labs to Legends program which educates the next generation of data scientists at universities on what ethical training for the music industry looks like and best practices to implement it
The music industry needs solutions to combat regulations and red tape that would slow down AI innovation. We need to be making it easier on AI companies that want to train ethical models rather than slowing them down. We need new frameworks to facilitate data set licensing and provide attribution.
Navigating Legal and Ethical Waters
This legislation fills a critical gap in copyright law, which hasn't kept up with AI's rapid advancements. Right now, AI-created works can’t be copyrighted under U.S. law, but using copyrighted material to train AI raises big legal and ethical questions. The bill aims to shed light on these practices, ensuring creators know how their work is being used and potentially getting compensated.
David Israelite, President & CEO of the National Music Publishers Association, put it:
“AI only works because it mines the work of millions of creators every day, and it is essential that AI companies reveal exactly what works are training their data.”
Joshua Landau, Computer & Communications Industry Association:
"The sheer scale does present practical issues," noting that models can train on more than 500 million works. Beyond identifying creators, that leaves little money per creator for even lucrative AI systems if royalties are eventually to be paid.
The bill arrives at a crucial time, with courts debating whether training AI on copyrighted materials constitutes infringement. Its requirement for detailed disclosure could set the stage for future legal and regulatory standards. Ashley Irwin, President of the Society of Composers & Lyricists, noted that this legislation is a significant step toward addressing the challenges AI poses to human-created copyrighted materials.
Wrapping It Up
In short, the Generative AI Copyright Disclosure Act is another bill in a long line of bills that focuses on transparency and copyright protection in AI training. While it's a hit with the creative crowd, it faces hurdles regarding practicality and its impact on innovation. It presents some new problems that lack definition. We will provide updates as we engage in ongoing communication with the committee regarding the various and unique challenges that the industry faces, particularly in relation to copyright issues and the complexities of proper attribution.
Read the Bill:
Related Updates
Questions?
Contact Us
© 2024 Sound Ethics Inc.
Visiting?
Finding Us
1800 Vine StreetHollywood, CA 90028, USA
Made by SF.CO
