Copyright law shifts for generative AI could restrict free tools

EFF warns narrow fair use rules for AI training risk blocking medical research and accessible tools

The Electronic Frontier Foundation filed an amicus brief in *In re OpenAI* arguing that restricting fair use for generative AI training could impede medical research and investigative journalism. The brief highlights how journalists and researchers use web scraping to uncover anticompetitive practices and misinformation, while tools like Be My AI rely on OpenAI’s models to provide visual descriptions for people with blindness. Professor Lawrence Lessig noted mandatory licensing for training data would disproportionately favor large tech companies. Courts have previously upheld fair use in cases where copyrighted works trained generative models, but the EFF warns that narrowing this scope could raise costs for developers and restrict access to tools that serve public needs.

This friction threatens the availability of free generative tools that support health, accessibility, and knowledge. If fair use protections are weakened, developers and researchers face higher barriers to creating affordable solutions—especially for marginalized groups who depend on low-cost AI tools. The EFF stresses that routine internet activities like streaming videos involve billions of untracked content copies, but the fair use findings apply specifically to training scenarios, not all AI use.

The source provides limited detail on implementation risks and global impact. A reader must consult the full amicus brief for context on how this affects specific jurisdictions or technical thresholds.

Source: EFF Deeplinks