WASHINGTON—Critics argue that developers of generative artificial intelligence (AI) systems are unfairly training their models on copyrighted works, but a new report from the Center for Data Innovation concludes that concern, among other frequent arguments about how generative AI is unfair to creators, is misguided. However, there are legitimate intellectual property (IP) rights at stake that deserve policymakers’ attention.
“Generative AI models should be allowed to train on legally accessed copyrighted content. Attempts to stop training these models merely reflect the latest AI techno panic,” said Daniel Castro, director of the Center for Data Innovation who authored the report. “Generative AI systems should not be exempt from complying with IP laws, but neither should they be held to a higher standard than human creators.”
The report examines common arguments for why generative AI is unfair to creators and breaks down why each lacks merit:
| Common Criticism | Response |
| Training generative AI systems on copyrighted content is theft. | Seeking inspiration and learning from others is not theft. Calling this process theft is clearly inaccurate when applied to the way humans observe and learn, and it is equally inaccurate to describe training a generative AI system. |
| Generative AI systems should not be allowed to train on content without the copyright owner’s explicit permission. | People are free to observe copyrighted works and use what they learn from them to create future content without the explicit permission of the copyright owners. There is no intrinsic rationale for why users of generative AI systems would need to obtain permission to train on copyrighted content they have legal access to. |
| Generative AI systems should compensate copyright owners for training on their content. | Copyright owners do not have this same expectation when other human creators learn from their works. Critics of generative AI are also likely to overestimate individual contributions. The size of generative AI models are huge, and the contribution of any single person is miniscule. |
| Generative AI systems should be prohibited from producing content based on the style of an artist without their permission. | It is perfectly legal to commission someone to write an original poem in the style of Dr. Seuss or an original song in the style of Louis Armstrong. Users of generative AI systems should retain the same freedom. |
| Generative AI systems should not be allowed to use fragments of copyrighted content in their outputs. | This argument reflects a poor understanding of how generative AI systems work. Generative AI systems do not produce remixes of existing content. They are not taking small samples of various works, altering them, and then recombining them in a new order. Instead, generative AI systems use massive amounts of training data to create incredibly complex prediction models that allow them to produce realistic content in response to specific prompts. |
The report acknowledges and explores legitimate IP rights that are at stake in the wake of generative AI’s surge in popularity. The center urges policymakers to focus on five harmful activities:
- Infringing on copyrights of AI-generated works
- Distributing copyrighted content
- Creating forgeries
- Creating infringing content
- Impersonating individuals
“Restricting generative AI models from training on lawfully accessed content would significantly curtail the development and adoption of this technology,” said Castro. “Policymakers, instead, need to focus on strengthening other IP rights to protect creators.”
