Leaked files suggest Suno scraped millions of tracks and lyrics

Steelman: both sides, in good faith

The strongest fair case each way — we don't pick a winner.

The case for

Artists and rights holders spent years developing their craft to create valuable intellectual property, which Suno then scraped—millions of copyrighted tracks from streaming platforms—without permission or compensation to train a competing commercial product. The company's failure to disclose this training methodology, only revealed through a security breach, combined with its subsequent claim to have used only 'publicly available' material, suggests deliberate concealment of practices it knew were legally and ethically questionable. Copyright protections exist specifically to ensure creators control commercial uses of their work and benefit from their labour.

The case against

Training large AI models necessarily requires vast datasets that do not exist under restrictive opt-in licensing arrangements; preventing companies from using publicly available material would concentrate development among wealthy incumbents and stifle innovation. Many jurisdictions recognise fair use for transformative purposes including research and development, and AI training arguably constitutes such fair use rather than derivative reproduction. Current copyright frameworks predate digital distribution and artificial intelligence; overly restrictive prohibitions on training-data collection are anachronistic and would harm technological progress and consumer choice.