infoJustice

The House Judiciary Subcommittee on Courts, Intellectual Property, and the Internet examined the standards and policy considerations that should be applied to whether IP protection should be given to inventions (patents) or creative works (copyrights) generated with the assistance of artificial intelligence, including whether current and proposed rules on inventorship and authorship need to be changed.

Witnesses: 

• Claire Laporte: IP Fellow and Former Head of Intellectual Property, Ginkgo Bioworks, Inc.
• Joshua Landau: Sr. Counsel for Innovation Policy, Computer and Comms Industry Ass’n
• Sandra Aistars: Clinical Professor, George Mason University Antonin Scalia Law School
• Kristelia Garcia: Professor of Law, Georgetown University Law Center

Witnesses generally agreed that current laws of inventorship and authorship appropriately address most issues raised by generative AI. They argued that additional regulations are unnecessary with respect to copyright law, and would overcomplicate the patent process while disincentivizing innovation. Furthermore, they said that guidances crafted by the Copyright Office and US Patent and Trademark Office have been unclear if not ill-conceived, and might burden authors and/or inventors who use AI assistance.

• Aistars and Laporte argued that requiring authors and inventors to disclose details regarding AI assistance and/or proprietary info related to AI tools would be burdensome, and shouldn’t be necessary to gain IP protections. However, Rep. Issa pushed back by arguing that disclosure of how AI is used in the creative or inventive process is a legitimate purpose of the IP system.
• The witnesses’ belief that courts can use current law to address issues raised by generative AI was challenged by Rep. Ivey, who expressed concern about allowing courts to legislate.
• Rep. Schiff said he is already working on regulations which would, at minimum, inform rights holders when their works are being used to train generative AI systems. He compared his own draft proposal to the EU’s AI Act.

Witnesses advocated FOR extending IP protection to works that use generative AI as part of the inventive or creative process, but AGAINST IP protections for works that are substantially or completely generated by AI. A distinction was drawn between the use of AI for inventions (patents) and creative works (copyrights), and the witnesses argued that as long as there is some level of human control, intervention, or influence, IP protections can plausibly be claimed.

• Aistars and Laporte claimed that tracing and reporting every step at which AI is used is costly and generally impractical, which was met with little disagreement.

• Rep. Kiley noted that inventions (unlike artistic works) are measured against their usefulness, and Laporte used this distinction to further her argument that the use of modern tools shouldn’t render useful inventions any less patentable.
• Although the witnesses agreed that the partial use of AI should not render inventions or creative works unprotectable, they struggled to address the issues specifically raised by creative works. Aistars described a test for creative works based on human creativity and a “creative spark,” which Rep. Issa characterized as a maybe-too-subjective test.

Witnesses argued that overregulation could stifle American competitiveness in the AI and IP space. They argued that the focus of our IP law should be to encourage innovation and to create valuable technologies, which requires a generally low-regulation and low-complexity approach to AI and IP. Concerns about the direct use of AI against American institutions were also raised.

• Rep. Nadler asked whether a narrow construction of creativity would stifle invention; Laporte strongly agreed that it would, and that there is a greater risk of stifling innovation than there is of AI being misused.
• Rep. Ross expressed a desire to support American innovation in AI and emphasized the need for international competitiveness in this field.
• Rep. Cline asked to what extent China’s AI development affects the US patent system, and suggested that China could use AI to review our patents and challenge them en masse, but Laporte responded that this isn’t an issue, and is unlikely to become an issue given the challenges and costs associated with raising challenges in the PTAB.
• Landau discussed the potential for AI to have a direct effect on the IP system by allowing foreign adversaries to flood our patent offices with illegitimate filings, and suggested that simple attestations as to whether AI was used in crafting future filings could help prevent this.

While clear improvements to IP law were suggested with respect to remedies (switching from injunctive to monetary damages), the vision for a system of remuneration remained elusive. Although everybody conceptually agreed that existing rights holders should be compensated in some way for the use of their works, nobody offered detailed ideas as to how this could or should be accomplished. Only ex post compensation in the form of monetary damages for infringing uses was suggested.

• Rep. Lofgren and Rep. Issa were concerned about remedies and remuneration, with Lofgren being worried about the potential for injunctions to stop AI companies in their tracks, and Issa being unsettled by the lack of compensation for existing rights holders whose works are being used to train AI systems. 
• Garcia agreed with Rep. Lofgren that the use of a “liability rule,” or a remedy aimed at monetary damages, makes a lot more sense with respect to generative AI. Switching from injunctive to monetary damages, specifically in some AI-related cases of infringement, makes far more sense in terms of how AI companies use others’ works to train their systems, and with respect to American competitiveness in the AI space.
• The witnesses agreed with Rep. Issa that existing rights holders should be compensated, but offered no clear vision as to how a system of remuneration might look or operate.