AI firms' fair-use defense shaky as their own people call training 'astonishing theft'

rvp · x · 2026-09-25

The Decoder highlights the tension in AI companies' copyright position: while they defend model training under fair use, their own insiders have described the practice as "astonishing theft." That framing could undermine the fair-use rationale at the heart of ongoing disputes over training data.

Original post →

More from Safety

Safety channel →