Lawmakers Introduce AI Transparency Bill to Protect Consumers, Copyright Holders

Reps. Don Beyer and Anna Eshoo introduced a bill in late December to shed light on the data used to train artificial intelligence foundation models.

A summary indicates that the AI Foundation Model Transparency Act would require companies responsible for such systems to disclose “whether user data is collected in inference” and whether any copyrighted material was used, Nextgov/FCW reported Tuesday.

The bill would also task AI firms to collaborate with the Federal Trade Commission, the National Institute of Standards and Technology and the Office of Science and Technology Policy to craft transparency standards. 

Beyer, the Congressional AI Caucus’ vice chair, stated that the AI Foundation Model Transparency Act would shed light on how a given system was built and what information it sources from. He explained that it is typically difficult to understand why AI foundation models produce a certain response.

Caucus co-chair Eshoo said the bill would enable consumers to make informed decisions when interacting with AI.

News of the AI Foundation Model Transparency Act comes after lawmakers put forward similar AI transparency measures during the previous Congress.

In September, two senators asked the OSTP to work on standards for authenticating AI-generated works to promote public trust.