What’s Covered?
This volume, edited by Brent Lutes, Chief Economist at the U.S. Copyright Office, sets the stage for a broad research agenda about the economic implications of AI for copyright policy. It’s not a legal analysis, nor does it stake out policy positions. Rather, it lays out which economic questions matter most, and how future research can better inform policymaking.
The report is divided into eight substantive parts:
- Introduction (Lutes) – Provides the framework: copyright’s economic logic is about balancing incentives to create with access to creative works. AI complicates both sides of that equation.
- Copyrightability and Demand Displacement (Jaffe, Reimers, Waldfogel) – Examines how AI-generated content competes with human creators and potentially reduces the value of human-made works.
- Infringement by AI Output (Smith, Telang) – Focuses on the economic consequences of AI output that mimics or replicates copyrighted material.
- Commercial Exploitation of Name, Image, and Likeness (Tucker) – Discusses how generative AI affects rights of publicity and the economic value of individual identity.
- Creative Incentives and Ingestion (Gans, Nagaraj) – Looks at how the ingestion of copyrighted works into training sets affects incentives to create.
- Developers’ Access to Training Data (Greenstein) – Explores the importance of training data access for innovation and the risk of market concentration.
- Controlling Training Inputs (Nagaraj) – Considers technical and policy approaches to limiting what data AI models can ingest.
- Socioeconomic Biases (Palmedo, Safner) – Raises equity and access questions, particularly for underrepresented groups affected by AI policy.
A key concept introduced is the distinction between “bounded” and “unbounded” AI models:
- Bounded = trained on clearly licensed or controlled data (e.g., a known set of books).
- Unbounded = trained on massive scraped datasets where rightsholders can’t be easily identified.
Most of the analysis focuses on unbounded models, since they raise the thorniest policy and market design problems.
The report highlights how generative AI changes the cost structure of content creation (by slashing marginal costs), introduces new market dynamics (substitute vs. complementary AI content), and challenges the idea that copyright alone can balance incentives and access. It also calls attention to foundation models as a central concern—due to their scale, opacity, and economic leverage.
💡 Why it matters?
This is the most comprehensive public-sector framing to date of how copyright policy needs to adapt to AI. It moves beyond yes/no questions about legality and reframes the issues around incentive structures, market failures, and the economics of creativity. It encourages a new generation of evidence-based policymaking—especially in an area where emotions, headlines, and litigation often crowd out analysis.
What’s Missing?
This report doesn’t aim to produce policy guidance or propose interventions. That’s by design—but it also leaves a gap for policymakers looking for ready-to-use tools or models. For instance:
- There’s no economic modeling of compensation mechanisms for training data use.
- There’s no real discussion of open source AI and its distinctive economics.
- It doesn’t cover international trade-offs, like cross-border copyright enforcement or differences in data mining exceptions.
Also, while the report talks a lot about foundation models, it doesn’t explore what happens downstream, where small developers and users face their own risks and costs.
The result is a framework that’s rich but not yet operational. The next step is turning these questions into empirical studies, simulations, and market data.
Best For:
Academic economists, policy researchers, think tanks, and anyone tasked with building copyright policy in an AI-saturated world. Also useful for non-lawyer AI governance professionals who want to understand how copyright debates are shifting.
Source Details:
Title: Identifying the Economic Implications of Artificial Intelligence for Copyright Policy: Context and Direction for Economic Research
Editor: Brent Lutes, Chief Economist, U.S. Copyright Office
Published: February 2025
Contributors: Joshua Gans (Toronto), Shane Greenstein (Harvard), Adam Jaffe (Brandeis), Imke Reimers (Cornell), Michael Smith & Rahul Telang (CMU), Catherine Tucker (MIT), Joel Waldfogel (Minnesota), and others
Institutional Context: This report emerged from a U.S. Copyright Office roundtable convened to explore economic—not legal—dimensions of AI and copyright. It is part of the Office’s effort to develop an empirical research base for future policymaking.
