The following note is from our bi-weekly Re:Create Recap click here to subscribe.

This week saw the release of the Bipartisan House Task Force Report on Artificial Intelligence, an in-depth exploration of a wide range of policy issues raised by AI technology. The Report’s discussion of intellectual property ably captures the nuances in this complex area and recognizes that, “It will be vital to avoid overreach” in government intervention in IP.

One of the Report’s “Key Findings” is that “Due to the rapidly shifting landscape and tremendous uncertainty of AI’s impacts on IP, Congress should exercise caution when considering legislation related to AI and IP.” The Report makes clear that the courts are the proper venue for sorting out the question of copyright’s application to AI training, for example; this issue is “not yet ripe” for government intervention.

The Report includes a kind of litmus test to determine whether IP proposals are reasonable or overreaching:

Any new IP-related legislation or regulations should target specific known issues or problems; tailor definitions, requirements, and consequences narrowly; reduce uncertainty rather than increase it; and focus on improving the ability of the private sector to innovate and creators to thrive.

Every major bill proposed this year on IP and AI fails this test miserably. NO FAKES, for example, is anything but “targeted”—it creates an entirely new class of federal intellectual property and would regulate every use of digital replicas outside a narrow set of carveouts. The requirements and consequences of the bill are thus extraordinarily broad; it implicates virtually every form of digital speech and every online speech platform. And NO FAKES absolutely increases uncertainty; as the EFF has shown, the untested boundaries of NO FAKES are a plaintiff’s lawyer’s dream come true—and a nightmare for the rest of us. Most importantly, the focus of NO FAKES is not enabling innovation or creativity, but rather adding a layer of uncertainty and doubt to the use of increasingly ubiquitous digital tools. Far from a modest, targeted intervention to address a specific known issue, NO FAKES is a strikingly broad bill whose full implications are impossible to predict.

The same could be said of every IP-related AI bill introduced in Congress this year, including the COPIED Act, the TRAIN Act, and PADRA. These bills leapfrog the courts and create broad, unpredictable new obligations for online creators and the platforms, services, and tools on which those creators rely. While the Report does recommend addressing the harms of “deepfakes,” it’s clear that NO FAKES is not the right approach. It’s time for Congress to abandon wildly overbroad approaches like NO FAKES and look for more tailored, less disruptive solutions to specific, pressing challenges arising from AI technology. The House Task Force Report effectively presses the Reset Button on IP and AI.