Federal courts are issuing the first substantive rulings on whether training AI models on copyrighted content constitutes fair use, setting precedents that will shape the industry for decades.
Key Cases
- NYT v. OpenAI: Court ruled some training uses may not be fair use
- Getty v. Stability AI: Partial summary judgment favoring Getty
- Authors Guild v. Meta: Case proceeding to trial in September 2026
- UMG v. Suno/Udio: Music generation copyright case fast-tracked
The Legal Framework
Courts are applying the traditional four-factor fair use test, with particular focus on whether AI outputs compete with the original works in the marketplace. Transformative use arguments have had mixed results.
Industry Impact
AI companies are increasingly licensing training data proactively. OpenAI signed deals worth $500M+ with publishers. Anthropic, Google, and Meta are building licensed data partnerships. The era of free training data may be ending.
