Following the release of the White House’s “America’s AI Action Plan,” organizations across various industry sectors commented on the plan’s impact on their efforts. Although the plan did not directly address copyright and AI, the President stated in his speech at last week’s AI Summit that additional copyright regulations could hinder America’s innovation agenda, and that the current copyright system is sufficient to protect both copyrighted works and the training of AI models.
In a blog post last week, Executive Director Brandon Butler broke down what the plan means for fair use.
Butler writes,
“Fair use is a principle that is at the heart of America’s balanced copyright system. While not all Re:Create members agree on every part of the AI Action plan, there is a common theme throughout our members’ comments to the Administration: let fair use be.
We applaud the Administration for doing just that: for prioritizing American innovation by discouraging government overreach in AI development. Re:Create remains committed to ensuring that copyright functions as a force for innovation and creativity, not as a source of red tape and barriers.”
A full roundup of Re:Create members’ comments to the AI Action Plan can be found below.
Foundation, request for input on the development of an Artificial Intelligence (“AI”) Action Plan directed towards sustaining and enhancing America’s AI dominance. We believe the highest priority action items relate to two dimensions of ensuring that copyright law does not become a barrier to the development of AI:
- First, the courts should treat the ingestion of content necessary to power many AI activities as a fair use under 17 U.S.C. § 107. To that end, the Action Plan should call for the Administration to file amicus briefs in support of fair use in appropriate cases.
- Second, license terms purporting to abridge the fair use rights of users should not be enforced. This can be achieved under theories such as conflict preemption advocated by the Administration in amicus briefs in litigation.
While copyright can provide a reward for creativity, an overbroad conception of copyright and its over-enforcement is fatal to creativity. Balanced copyright laws, including fair use and text-and-data mining (TDM) exceptions, provide a legal framework that supports innovation, free speech, the public interest, and the rights of content creators alike….
The Administration can best support AI by actively advocating for these exceptions, both at home and abroad in the form of fair use or TDM exceptions. This advocacy can include regulatory policy work, direct engagement in foreign jurisdictions, and amicus support in the courts.
The AI Action Plan should adopt balanced copyright rules that protect rights holders while enabling AI systems to learn from prior knowledge and publicly available information. Specifically, the Administration should ensure access to publicly available scientific papers to accelerate AI in science fields. The Administration should establish that the fair use doctrine and text and data mining exceptions allow for uses of copyrighted, publicly available material for AI training purposes, so long as rights holders can opt out of such use. This mechanism will avoid significant impacts on rightsholders while avoiding unpredictable and lengthy negotiations with data holders during model development.
Some have suggested that licensing schemes provide a way to address creators concerns about use of their works for training. But any such scheme carries significant risks. Requiring developers to license the materials needed to create AI technology threatens the development of more innovative and inclusive AI models, as well as important uses of AI as a tool for expression and scientific research. Specifically, requiring AI developers to get authorization from rights holders before training models on copyrighted works would make it harder for newer companies that don’t have their own trove of training data to create new tools. Instead, this scheme benefits giant tech monopolists that can afford to pay pricey licensing deals that lock in their dominant positions in the generative AI market by creating prohibitive barriers to entry.
To ensure an open and competitive AI sector, we propose four key policy priorities:
- Protect the rights to read and learn that make AI training possible.
- Support open-source, collaborative research and development.
- Build and maintain public physical and digital AI infrastructure to prevent monopolization and private enclosure.
- Develop standards and sensible rules around AI explainability and transparency to ensure trust and adoption.
Fair use is crucial in promoting innovation and creativity in various fields including education, research, journalism, and criticism. These principles must be part of any attempt to apply copyright to issues surrounding AI. For example, AI training on copyrighted works should qualify as non-infringing or fair use because AI is learning ideas rather than reproducing specific expressions. Attempting to establish a licensing framework for AI training data would create significant economic barriers, stifle innovation, and potentially lead to market monopolization by large tech companies with the resources required to operate under a licensing regime.
###
About Re:Create: Re:Create is a coalition comprised of a broad membership of think tanks, advocacy organizations, libraries, technology companies – large and small – that serves as the leading coalition united in the fight for a balanced copyright system that is pro-innovation, pro-creator, and pro-consumer. Not every member of the Re:Create Coalition necessarily agrees on every issue, but the views we express represent the consensus among the bulk of our membership.