Copyright and Artificial Intelligence Part 3: Generative AI Training Report Pre-Publication Version 2025

• Authority: Copyright Office, Library of Congress
• Parent authority: Federal government
• Jurisdiction: United States
• Collections: Miscellaneous documents
• Most recent activity: Proposed
• Most recent activity date: 2025-05-09

Risk subdomain coverage

Minimal coverage — Brief mention but no specific mitigations, recommendations or governance measures relating to the risk subdomain are described.

• Power centralization and unfair distribution
• Economic and cultural devaluation of human effort
• Competitive dynamics
• Governance failure

All other risk subdomains had no coverage in this document.

Risk Coverage Summary

The document has minimal coverage of risk domains, with limited focus on competitive dynamics (6.4) and governance failure (6.5). The document primarily addresses copyright licensing mechanisms rather than AI risks, with brief mentions of market concentration concerns and regulatory framework considerations. Most risk subdomains receive no coverage.

Sector coverage

Good coverage — Comprehensive governance measures or mitigations with clear procedures relating to the sector are explicitly described.

• Information
• Arts, Entertainment & Recreation

Minimal coverage — Brief mention but no specific mitigations, recommendations or governance measures relating to the sector are described.

• Professional & Technical Services
• Scientific Research & Development Services
• Public Administration (excl. National Security)

All other sectors had no coverage in this document.

Sectors Summary

The document governs AI training across multiple creative and information sectors, with explicit coverage of Arts, Entertainment, and Recreation (music, photography) and Information (AI development, data processing). It also addresses Professional and Technical Services through licensing and legal considerations, and Scientific Research through academic research exemptions.

Legislative Status

Soft Law

Legislative Status Summary

This is a non-binding recommendation report from the Copyright Office that uses voluntary language and explicitly recommends against government intervention, favoring market-based solutions.

Actors

Proposer

• Entity Types: AI_Governance_Actor
• Entities: Copyright Office, Library of Congress
• Summary: The document is authored by the Copyright Office, Library of Congress, as indicated in the title and authority field. This is a government agency providing recommendations on AI training and copyright licensing.

Target

• Entity Types: AI_Developer, AI_Governance_Actor
• Entities: AI companies; copyright owners; creators; Collective Management Organizations (CMOs); academic researchers; smaller developers; Department of Justice
• Summary: The document addresses AI developers who train models on copyrighted works, copyright owners and CMOs who license content, and recommends DOJ provide antitrust guidance. It discusses licensing arrangements between these parties and considerations for different types of developers.

Enforcer

• Summary: The document does not establish enforcement mechanisms or designate enforcement authorities. It explicitly recommends against government intervention and favors voluntary market-based licensing solutions.

Monitor

• Entity Types: AI_Governance_Actor
• Entities: Copyright Office; courts
• Summary: The Copyright Office implicitly positions itself as monitoring market developments and legal clarity, noting the need to observe how licensing markets evolve. Courts are mentioned as resolving pending cases that will provide legal clarity.

AI Lifecycle Stages

Lifecycle stages covered by the documents, as defined in the OECD Framework for the Classification of AI Systems

• Collect and Process Data
• Build and Use Model

AI Lifecycle Summary

The document primarily addresses the 'Collect and Process Data' stage of AI development, focusing extensively on licensing copyrighted works for use in AI training datasets. It also touches on 'Build and Use Model' through discussions of training AI systems and model architectures.

Technical Scope

Terms relating to technical scope that are mentioned or implicitly referenced in the document

• AI Models
• AI Systems
• Generative AI
• Task-specific AI

Technical Scope Summary

The document explicitly mentions AI systems and AI training, with references to LLMs and generative models. It does not use technical terminology like 'frontier AI', 'GPAI', or compute thresholds, focusing instead on copyright and licensing aspects of AI development.