This bill, known as the Transparency and Responsibility for Artificial Intelligence Networks Act (TRAIN Act) , establishes a new administrative subpoena process within copyright law. Its primary goal is to empower copyright owners to discover whether their protected works have been utilized in the training datasets of artificial intelligence models, particularly generative AI models. Under this process, a copyright owner or their authorized agent can request a subpoena from a U.S. district court clerk. The request must include a proposed subpoena and a sworn declaration affirming a good faith belief that the developer used their copyrighted works for training. The subpoena is strictly limited to seeking information about the requester's own copyrighted materials. Upon receiving a properly filed request, the clerk is mandated to expeditiously issue the subpoena to the AI model developer. The developer is then required to promptly disclose copies or records sufficient to identify the copyrighted training materials. The bill defines "developer" as entities that design, own, or substantially modify generative AI models and curate or use training datasets, excluding noncommercial end users. Recipients of these records are bound by a duty of confidentiality , preventing unauthorized disclosure. The Federal Rules of Civil Procedure govern the subpoena's issuance, service, and enforcement. Notably, a developer's failure to comply with a subpoena creates a rebuttable presumption that they copied the copyrighted work. Conversely, courts can impose sanctions on copyright owners who make bad faith subpoena requests, aligning with Rule 11(c) of the Federal Rules of Civil Procedure.