What Past Search Engine Battles Teach Us About Fair Use and GenAI

Courts have repeatedly rejected claims that technologies analyzing copyrighted works threaten creativity, instead treating such analysis as lawful and socially beneficial.

In an article from the Electronic Frontier Foundation, Joe Mullin writes about how search engines, an early flashpoint, were accused of mass infringement for copying content to index and retrieve information. But judicial decisions concluded that copying for understanding, indexing, and locating information is consistent with fair use and essential to an open internet.

The same legal question now frames disputes over fair use and GenAI and whether copyright owners control downstream analysis of existing works.

The historical backdrop reflects a consistent pattern. Copyright holders challenged photocopiers, VCRs, and later internet search tools, arguing that unlicensed copying would undermine markets. Courts distinguished analytical copying from substitutive use, finding that technologies enabling search, research, and learning transform works rather than replace them.

That reasoning allowed libraries, researchers, and technologists to analyze large bodies of text without securing licenses from innumerable rights-holders, preserving copyright’s balance between protection and access.

Applying those principles to AI, courts have recognized that model training involves studying patterns across works, not reproducing expressive content. In Bartz v. Anthropic, a court concluded that AI training is highly transformative, and that alleged market harm was speculative. The decision focused on whether training is non-substitutive and rejected abstract theories that AI outputs inherently compete with original works.

The analysis cautioned against expanding copyright to restrict learning itself, which could destabilize established practices in science, medicine, and data-driven research.

The ongoing debate will affect the legal aspects of technology transactions, licensing strategy, and due diligence regarding data use in AI development. Counsel should assess fair use risk alongside contractual restrictions, regulatory exposure, and cross-border data transfer rules.

Expansive licensing mandates may raise antitrust and governance concerns. Advising boards on disclosure, enterprise risk management, and intellectual property strategy requires understanding how fair use limits copyright control and shapes enforcement, litigation, and deal structuring in AI-driven markets.