The Electronic Frontier Foundation (EFF) has filed an amicus brief in the In re OpenAI litigation, arguing that a ruling against fair use in generative AI training would have catastrophic side effects for the entire internet ecosystem. The brief, filed on October 30, 2026, contends that the plaintiffs' demand for stricter copyright controls ignores the fundamental mechanics of how data is acquired online. If courts accept the argument that scraping or downloading data without explicit permission constitutes infringement, the legal foundation for routine web interaction collapses.

The Technical Reality of 'Intermediate Copying'

EFF’s legal team highlights a critical infrastructure issue: the requirement that users prove copyrighted material was lawfully obtained would cast legal doubt on billions of daily copies made during routine online activities. As detailed in the brief, actions such as opening emails, clicking embedded links, visiting websites, and streaming music all require intermediate copying. Fair use currently acts as the safety valve that allows these processes to function without users needing to track and clear rights for every single image, video, or article they encounter. Removing this protection would effectively mandate a permissions-based internet, a model that is technically unfeasible for most developers and users.

Impact on Research and Development Tools

The brief draws a direct parallel between AI training methods and the data acquisition techniques used by journalists, researchers, and watchdog groups. These entities often use 'crawling' or 'scraping' to investigate anticompetitive practices by tech giants or to analyze misinformation on platforms like X. EFF argues that if the plaintiffs' theory prevails, these non-profit and academic efforts would face the same legal hurdles as AI companies, but without the financial resources to litigate. Furthermore, the argument that AI training is unfair because it might produce competing works ignores the historical purpose of copyright, which is to stimulate new creation rather than protect incumbents from competition.

Licensing Models Favor Big Tech Monopolies

Proponents of mandatory licensing for AI training argue it would compensate creators, but EFF and Professor Lawrence Lessig suggest it would actually entrench the power of existing tech monopolies. Lessig noted that if all training requires a license, the inefficiency of the copyright system means only a handful of tech giants could afford the costs. This creates a barrier to entry for smaller developers and open-source projects. Additionally, EFF points out that media conglomerates have historically exploited the imbalance between themselves and individual artists, meaning that granting more rights to copyright holders often benefits the publishers rather than the creators they claim to protect.

Key Takeaways

  • Narrowing fair use in AI cases could legally jeopardize routine web activities like streaming and email due to 'intermediate copying' requirements.
  • Scraping tools used by journalists and researchers face the same legal risks as AI trainers, potentially stifling investigative tech journalism.
  • Mandatory licensing schemes may benefit Big Tech monopolies that can afford compliance costs while excluding smaller developers and open-source projects.
  • The EFF brief emphasizes that copyright law should serve the public good, not just resolve disputes between media conglomerates and AI giants.

The Bottom Line

Restricting fair use to appease copyright holders risks breaking the fundamental plumbing of the internet, ultimately serving only Big Tech monopolies and media conglomerates at the expense of developers, researchers, and everyday users.