AI-generated art can’t be copyrighted after Supreme Court declines review

Original Article ↗ Hacker News Discussion ↗

Scope of the Ruling & Human Authorship

  • Discussion emphasizes that this case is narrow: it rejected a registration where the filer explicitly named the AI as the author and denied human authorship.
  • Key legal concepts highlighted:
    • Copyright requires a human author.
    • There must be minimal creativity, not purely mechanical output.
    • Work must be fixed in a tangible medium.
  • Several commenters stress that the decision does not fully resolve how much human input is needed to claim authorship over AI-assisted works.

AI vs Photography and Other Tools

  • Frequent comparison: camera vs generative model.
    • Photography: human controls composition, timing, lens, etc., so expressive choices are clearly human.
    • AI image generation: user gives high-level prompts; the system determines composition and details, so authorship is less direct.
  • Some argue that heavy prompting, iterative refinement, and post-editing should qualify as “substantial human authorship.”
  • Others counter that a prompt is more like a commission or instruction than the artwork itself, and AI is not “just another brush.”

AI-Assisted vs Fully AI-Generated Works

  • General thread consensus:
    • AI-assisted works can be copyrightable if human-authored elements are perceptible (e.g., edits, arrangement, integration into larger human-made works).
    • Pure AI output, with the system determining expressive content, is likely not copyrightable under current US guidance.
  • Edge cases discussed: small edits (e.g., adding one pixel) likely do not meet “substantial human authorship.”

Code, Software, and Other Outputs

  • Unclear whether fully AI-generated code is copyrightable; no major case yet.
  • Some hope AI-generated code is not protected, arguing symmetry with training on copyrighted code.
  • Others note that much real-world code is AI-assisted, reviewed, and modified, which likely restores human authorship for significant portions.
  • For proprietary backends, some say copyright matters less than trade secret and access control.

Fairness, Policy, and Future of IP

  • Some argue: if training on copyrighted works is allowed, it is fair that outputs lack new copyright.
  • Others warn that if everything becomes AI-generated, copyright’s role diminishes, and patents or trade secrets may become more important.
  • There is both enthusiasm for AI as a new artistic medium and strong skepticism that prompting can ever match human-led craftsmanship.