Coalition of AI Developers Issues New Report on AI Training and Copyright
Leading IP Attorneys, AI Progress Coalition Clarify How AI Training Works – and Why U.S. Copyright Law Supports It
Washington, D.C. — As debates over generative artificial intelligence (AI) training and copyright continue, AI Progress today released a new report, AI Models: Addressing Misconceptions About Training and Copyright, which sets the record straight on how AI models learn, what U.S. copyright law protects, and why training AI models on copyrighted works is consistent with decades of legal precedent under the fair use doctrine.
The report comes from AI Progress, a coalition that includes Amazon, Anthropic, Cohere, Google, Meta, Midjourney, Microsoft, and OpenAI. It is authored by Anna Chauvet, AI Progress advisor and partner and copyright practice leader at Finnegan, and Karthik Kumar, Ph.D., a partner at Finnegan specializing in IP litigation and emerging technologies. Together, they call attention to the risks of misinterpreting copyright law in ways that could undermine U.S. competitiveness.
“Generative AI models learn patterns and relationships from large, diverse datasets to generate new, original outputs,” Chauvet said. “U.S. copyright law has long protected creativity while allowing new technologies to flourish. Far from a departure, recent court rulings align with decades of legal precedent: exceedingly transformative uses — like AI model training — are a fair use.”
Key takeaways from the report include:
- AI models generate new works based on learned patterns — not by retrieving stored data.
- Copyright law protects creative expression but not facts or ideas in original works.
- Courts have held that training AI models on copyrighted works is a transformative fair use.
- Efforts to restrict AI training through changes to existing U.S. copyright law threaten innovation.
“America has led every major technology wave because our copyright framework has encouraged innovation,” Kumar said. “The same legal principles that enabled past breakthroughs must continue to guide AI development.”
The full report, AI Models: Addressing Misconceptions About Training and Copyright, and summary fact sheets are available here.
###
About AI Progress
The AI Progress coalition is dedicated to the responsible development of artificial intelligence, focused on delivering benefits across sectors and society and ensuring the U.S. stays competitive by supporting policies that encourage innovation and protect creativity. Members include Amazon, Anthropic, Cohere, Google, Meta, Midjourney, Microsoft, and OpenAI. AI Progress is on X and LinkedIn.
