Analyses / Public Summary / 119 · HR 7209 Public Summary

119-HR-7209 Journalist Public Summary

119 · HR 7209 TRAIN Act

Creates a fast, clerk-issued subpoena so copyright owners can find out if their works were used to train generative AI models; supporters cite transparency for creators, while critics worry about burdens, confidentiality, and a presumption of copying if developers don’t comply.

Published
23 Jan 2026
Updated
23 Jan 2026
Tags
US Congress · 119th Congress · AI
Unvetted
01 · Section

Headline Summary

A bipartisan House bill would let copyright owners quickly obtain records from AI developers to learn whether specific copyrighted works were used to train generative AI models.

02 · Section

What It Does

The TRAIN Act (H.R. 7209) creates an administrative subpoena process in Title 17. A copyright owner (or authorized agent) can ask a federal district court clerk to issue a subpoena requiring an AI developer to disclose copies of, or records that identify with certainty, the owner’s works used as training material for a generative AI model. The requester must submit a sworn statement of a subjective good‑faith belief that their works were used, and the subpoena can cover only the requester’s own works. Once served, the developer must disclose the requested records expeditiously, and the recipient must keep the information confidential.

  • Process rules mirror federal subpoena procedures unless otherwise specified.
  • If the clerk finds the paperwork is in proper form, the clerk issues the subpoena without a judge’s discretionary review.
  • Noncompliance by a developer creates a rebuttable presumption that the developer made copies of the copyrighted work.
  • Courts may sanction requesters who act in bad faith, using the Rule 11 framework.
  • Takes effect upon enactment.
03 · Section

Who’s For It

  • Sponsors: Rep. Madeleine Dean (D‑PA) and Rep. Nathaniel Moran (R‑TX).
  • Supporters argue it gives creators a straightforward way to verify if their works trained AI systems, which they say is necessary to evaluate potential infringement and seek remedies if warranted.
  • Some arts, publishing, and media stakeholders are likely to view the bill as a transparency tool that doesn’t itself decide legality, but helps gather facts.
04 · Section

Who’s Against It

  • AI developers and tech firms may warn that broad disclosure demands could expose trade secrets or sensitive datasets, even with confidentiality rules.
  • Civil‑liberties and privacy advocates could question the low threshold (“subjective good‑faith belief”) and fear fishing expeditions or burdens on smaller labs.
  • Defense‑side lawyers may argue the rebuttable presumption for noncompliance effectively shifts leverage toward requesters, pressuring disclosures even when contested.
05 · Section

What’s Next

Status as of January 22, 2026: introduced in the House and referred to the Judiciary Committee. Next steps typically include committee hearings or markups, a House floor vote, Senate consideration, and then the President’s signature or veto.

Discussion