Congress proposes a bill requiring AI companies to report copyrighted training d

Small businesses and startups building generative AI products would need to catalog and report every copyrighted work in their training datasets to the Copyright Office before releasing their models. This creates a significant compliance burden — especially for smaller companies that may lack legal teams — but also levels the playing field by applying the same transparency rules to everyone. Companies face fines of at least $5,000 per unreported work, up to $2.5 million per year, plus potential injunctions halting their AI products.

Many gig workers are freelance writers, artists, musicians, and other creators whose copyrighted works may be used to train AI models. This bill would give them visibility into whether their registered copyrighted works are being used, and a legal avenue — including attorney's fees — to hold AI companies accountable if they fail to disclose that use. This doesn't stop AI companies from using the works, but it gives creators important transparency.