Cognixia USA’s Post

Recently, a new Generative AI Copyright Disclosure Act was proposed in Congress, which would mandate companies to disclose copyrighted training data to the US government at least 30 days before model launch or face fines. This addresses concerns about lack of transparency regarding training data, especially its potential infringement on copyrighted materials. If passed, it would give more power to content creators to negotiate their own terms.As this never ending debate continues, share your thoughts in the comments below. #GenAI

  • No alternative text description for this image

To view or add a comment, sign in

Explore topics