James Gibson (U Richmond Law) and Christopher Anthony Cotropia (George Washington U -Law) have posted “Slowing Down AI with IP” on SSRN. Here is the abstract:
The current artificial intelligence (AI) landscape is defined by “too much, too fast”: rapid, winner-take-all scaling by a handful of firms, backed by states and capital markets, with little transparency into training data, model architectures, or deployment practices. The resulting harms are widely recognized: environmental damage, labor displacement, economic instability, privacy violations, and perpetuation of bias.
What is not widely recognized is the role that intellectual property (IP) can play in mitigating these harms. Existing policy debates largely treat IP protection as an AI accelerant, framing patent and copyright as incentives for more and faster AI innovation. If this were true, granting IP rights to AI innovators would only make things worse. But this Article inverts the conventional framing. When it comes to AI development, IP’s incentive effect is minimal, whereas IP’s well-known costs—e.g., reduced production, slower diffusion, and mandatory disclosure—predominate.
Maintaining and expanding IP protection for AI technologies therefore has the opposite of its usual effect: it slows down innovation. This makes it a uniquely fitting regulatory tool for the “too much, too fast” AI landscape. And while IP is by no means a cure-all, it has unexpected advantages over more common regulatory strategies. Overall, we reposition IP not as fuel for the AI race, but as a necessary brake on a runaway industry.