US Department of Justice argues AI training on copyrighted works is transformative fair use

The US Department of Justice has intervened in multiple lawsuits involving OpenAI and Anthropic by filing a Statement of Interest. The government argues that the process of training models on written works is transformative, which would categorize the practice as fair use under legal doctrine. This position supports AI companies against a broad group of plaintiffs, including authors and publishers like Ziff Davis and The New York Times.

Adopting this fair use standard removes the cost of licensing fees for AI companies during the ingestion phase. However, this approach creates a significant cost for content creators who lose control over their intellectual property. The DOJ specifically rejects the notion that AI-generated works in the same genre as the original material cause market harm.

The government's stance is driven by the belief that limiting training data would compromise US competitiveness and national security. This creates a tension where the state prioritizes the rapid development of AI capabilities over traditional copyright protections. Publishers argue that this strategy may destroy the human-created content that AI requires to function.