Fair Use in Training AI Models

• On June 23, 2025, Judge Alsup of the U.S. District Court for the Northern District of California held that Anthropic’s use of copyrighted materials to train its AI models was fair use because the models’ purpose was “exceedingly” transformative. Two days later, Judge Chhabria, of the same court, ruled that Meta’s use of copyrighted materials to train its AI models was fair use because their purpose was “highly” transformative.
• Judge Alsup also held that the digitization of purchased materials for the purpose of building a searchable “central library” was fair use, while the collection of reproduced material for the same purpose was not.
• The decisions differ in their analyses under the fair use doctrine and are tied to the specific facts in their respective cases. Additionally, courts (including, eventually, the U.S. Supreme Court) may reach different conclusions. The decisions nonetheless represent a positive development for AI companies seeking clarity on these significant questions.

Read More