Fallthrough

The Vibes-Based Legal System

Feb 7, 2026
Steve Klabnik, software developer and long-time contributor on software, licensing, and copyright law, digs into whether AI output can be copyrighted. He separates training from output, traces software’s treatment under the 1976 Act, and warns that owning a style would choke culture. The conversation also probes licenses, how copyright is wielded, and whether LLMs learning equals infringement.
Ask episode
AI Snips
Chapters
Books
Transcript
Episode notes
INSIGHT

Human Authorship Is Central To Copyright

  • Copyright only applies to human-created works under current U.S. law, making AI-generated code legally murky.
  • The mouse/monkey photo case exemplifies courts requiring human authorship for copyrightity.
INSIGHT

Training vs Output: Two Different Legal Questions

  • Training models and model outputs are separate legal questions with different precedents and uncertainties.
  • Fair use and context can mean verbatim snippets from copyrighted works may not always be infringement.
INSIGHT

Why Software Became Copyrightable

  • The 1976 Copyright Act created special treatment for software after deliberation about whether programs were recipes or novels.
  • Law historically treated source code as copyrightable similar to literary works, not as mere functional instructions.
Get the Snipd Podcast app to discover more snips from this episode
Get the app