Frontier AI & Model Developers

DOJ backs OpenAI's fair-use defense in NYT copyright litigation

Source: IPWatchdog · Sep 3, 2026

The U.S. Department of Justice has filed a Statement of Interest in the consolidated OpenAI copyright litigation in the Southern District of New York, arguing that training large language models on copyrighted text qualifies as fair use. The filing frames the question partly in national-security and competitiveness terms — an unusual angle for a fair-use argument, and a signal that Washington sees AI-training-data law as adjacent to industrial policy, not just intellectual-property doctrine.

The filing is advisory; Judge Sidney Stein still decides the underlying case on its merits, and DOJ's position doesn't bind the court. But a Statement of Interest at this stage puts a thumb on the scale in a case a lot of the industry is watching as a bellwether for how broadly "fair use" will be read against AI training generally.

The framing worth tracking closely: DOJ's argument is specifically about the legality of training on already-published, copyrighted text — a different question from how a lab acquired that text, or what a model produces downstream. That's a meaningful distinction for anyone selling data into this market. A ruling that broad fair-use protection covers training doesn't settle whether a given dataset was properly acquired, consented to, or licensed — it just narrows one legal exposure among several.

For a business built on privacy-safe, fully consented, licensed interview data, this matters less as a risk than as a market signal: as the fair-use question around scraped text gets litigated, licensed and provenance-clean data doesn't need that defense in the first place. It's a good marker for how the value of "we can tell you exactly where this came from" is likely to keep going up, not down, as this case and others like it play out.

Key Points

• DOJ filed a Statement of Interest in In re: OpenAI Copyright Infringement Litigation (SDNY)
• Argues LLM training on copyrighted text is fair use, citing national-security and competitiveness concerns
• Advisory only — Judge Sidney Stein still rules on the merits
• Separates the training-use question from data-acquisition and output questions