TRAIN Act

The TRAIN Act would establish a new administrative subpoena process under copyright law to help copyright owners determine whether their works were used to train artificial intelligence models. Under the proposed law, copyright owners who have a good-faith belief that their copyrighted works were used in AI training could request a subpoena from a U.S. district court clerk. This subpoena would compel AI developers to disclose copies of training materials or records that identify the copyrighted works used, but only those belonging to the requesting copyright owner. The bill includes protections against misuse, such as sanctions for bad-faith requests and confidentiality requirements for disclosed information.

If enacted, this bill would give creators, publishers, authors, and other copyright holders a targeted tool to investigate whether their intellectual property was used without permission in AI model development. Currently, only a handful of states have enacted AI training data transparency laws, and they use different approaches. This federal mechanism would be specifically designed for copyright owners to protect their rights and potentially pursue legal action if unauthorized use is discovered. The bill is currently in committee and has not yet been voted on by the full House.