- Targeted stakeholdersGives copyright owners a formal mechanism to discover whether their works trained generative AI models.
- DevelopersIncreases transparency of training datasets used by developers of generative AI models.
- Targeted stakeholdersFacilitates enforcement actions and licensing negotiations by providing documentary evidence of training use.
TRAIN Act
Referred to the House Committee on the Judiciary.
This bill adds a new section to Title 17 creating a subpoena procedure allowingcopyright owners (or authorized agents) to obtain copies of, or records identifying, works used to train generative artificial intelligence models.
A requester files a proposed subpoena and sworn declaration with a district court clerk; if in proper form the clerk must issue it and the developer must expeditiously produce the requested copies or identifying records.
Recipients must keep produced materials confidential, failure to comply creates a rebuttable presumption of copying, and Rule 11 sanctions apply for bad‑faith requests.
Narrow statutory fix with clear stakeholders but faces organized industry resistance and legal complexity; passage plausible in amended form, uncertain as-is.
Relative to its intended legislative type, this bill establishes a new substantive right/obligation by adding a Title 17 subpoena mechanism for copyright owners to obtain copies or identifying records of materials used to train generative AI models. It defines key terms and sets out a basic issuance-and-disclosure process, confidentiality requirements, a rebuttable presumption for noncompliance, and a sanctions path tied to Rule 11.
Left emphasizes creator transparency and accountability for AI training
Who stands to gain, and who may push back.
- DevelopersImposes compliance, recordkeeping, and disclosure costs on AI developers, including small firms.
- Targeted stakeholdersRisks compelled disclosure of proprietary datasets, model details, or trade secrets to rights holders.
- Permitting processUses a subjective good faith standard that may permit broad or speculative subpoena requests.
Why the argument around this bill splits.
Left emphasizes creator transparency and accountability for AI training
Likely supportive because the bill strengthens creators’ ability to learn whether their copyrighted works were used in AI training.
It aligns with demands for transparency, accountability, and remedies for alleged unauthorized use of expressive works.
Cautiously favorable to transparency for rights holders, but concerned about trade secrets, burdens on developers, and due process.
Would favor narrowly tailored safeguards and clearer procedural limits before broad adoption.
Skeptical due to potential burdens on innovation, disclosure of proprietary datasets, and expansion of litigation leverage for rights holders.
Concerned it could chill AI development and impose compliance costs.
The path through Congress.
Reached or meaningfully advanced
Reached or meaningfully advanced
Still ahead
Still ahead
Still ahead
Narrow statutory fix with clear stakeholders but faces organized industry resistance and legal complexity; passage plausible in amended form, uncertain as-is.
- Breadth of 'developer' definition and which entities are covered
- Treatment and protection of trade secrets and proprietary datasets
Recent votes on the bill.
No vote history yet
The bill has not accumulated any surfaced votes yet.
Go deeper than the headline read.
Left emphasizes creator transparency and accountability for AI training
Narrow statutory fix with clear stakeholders but faces organized industry resistance and legal complexity; passage plausible in amended for…
Relative to its intended legislative type, this bill establishes a new substantive right/obligation by adding a Title 17 subpoena mechanism for copyright owners to obtain copies or identifying records of materials used…
Go beyond the headline summary with full stakeholder mapping, legislative design analysis, passage barriers, and lens-by-lens tradeoff breakdowns.