US lawmaker proposes a public database of all AI training material

Key Points:

  • The Generative AI Disclosure Act requires AI companies to disclose copyrighted works used in training datasets.
  • The bill would create a publicly available online database for all notices submitted.
  • Violations of the act could result in a civil penalty of at least $5,000.


US Representative Adam Schiff has introduced the Generative AI Disclosure Act, requiring AI companies to disclose copyrighted works in their training datasets to the Register of Copyrights. This bill enforces transparency by listing referenced works and providing a URL for creators to verify and potentially seek compensation. The act aims to respect creators’ rights and contributions in the AI landscape.


The proposed legislation is retroactive and applies to all existing and future AI systems, with violators facing a minimum civil penalty of $5,000. The bill garnered positive feedback from creative groups, emphasizing the importance of transparency and protection for creators against unauthorized use of copyrighted materials in AI training.


At a hearing on artificial intelligence and intellectual property, Rep. Darrell Issa praised the bill as a thoughtful step, signaling potential consideration by the House Judiciary Subcommittee on Courts, Intellectual Property, and the Internet. Schiff emphasized the bill’s role in informing creators about their works in AI datasets and expressed willingness to collaborate on advancing the legislation.


Meredith Stiehm of the Writers Guild of America West and other creative leaders supported the bill, stressing the necessity for greater transparency and safeguards in AI practices to shield creators’ rights. The proposal addresses concerns of unauthorized use of creative content in AI models and establishes clear recordkeeping guidelines to enforce creators’ rights effectively.


While awaiting resolution of complex AI copyright issues, including ongoing lawsuits involving AI companies such as OpenAI and Stability AI, a US Copyright Office initiative plans to issue guidance on deepfakes and AI-generated materials. The focus remains on digital replicas and copyrightability of works incorporating AI outputs, with training AI models on copyrighted materials not immediately prioritized.


The Copyright Office aims to finalize its reports by the fiscal year-end, potentially influencing court decisions but not binding them to its guidance. As legal battles unfold concerning AI models and copyrighted works, including disputes involving authors and news publications, the impact of forthcoming copyright guidance and legislations on creators and AI companies remains significant in shaping the evolving landscape of AI and intellectual property rights.



Prompt Engineering Guides



©2024 The Horizon