AI companies must disclose copyrighted training data—but can still use it
S. 3813 — CLEAR Act · Filed by Adam Schiff (D-CA) · 1 cosponsor · Introduced Feb 10, 2026 · Referred to committee
Your members of Congress
Enter a ZIP to see where your representative and both senators stood on this bill.
Looked up on this device — your ZIP is never stored on our servers.
What it does
This bill requires AI companies and developers to notify the U.S. Copyright Office within 30 days before releasing or commercially using any generative AI model that was trained on copyrighted works. The notice must list each copyrighted work used and provide a link to the training dataset if publicly available. Violations can result in civil penalties of $5,000 per instance (capped at $2.5 million per year), injunctions stopping the AI's use, and attorney's fees paid to copyright owners. The Copyright Office will maintain a public database of all notices.
Why we flagged it
The bill's core mechanism is a mandatory disclosure requirement—not a ban on AI training with copyrighted works, not a licensing scheme, and not a compensation framework. It creates a registry and enforcement pathway for copyright owners to challenge non-disclosure, but does not restrict AI development itself.
What the text implies
- The bill does not require consent or compensation from copyright owners—only notice. AI companies can still train on copyrighted works; they simply must disclose it. This may legitimize AI training on copyrighted material as long as the notice is filed.
- The $2.5 million annual penalty cap is a hard ceiling per person/entity, not per violation. A large AI firm could train multiple models on thousands of copyrighted works and face a maximum $2.5M penalty annually, potentially making the cost of non-compliance negligible relative to the value of unrestricted training.
- The public database of training datasets may expose creators' works and their locations to competitors and bad actors, creating a searchable index of copyrighted material used in AI training without the creators' consent to that indexing.
- The 30-day pre-release notice window is very short and may be impractical for large-scale AI development, creating a compliance trap for smaller developers while large firms with legal teams absorb the cost.
- The bill does not address the underlying question of whether training on copyrighted works without permission is lawful—it only requires disclosure if it happens. This leaves the fair-use question unresolved and may inadvertently codify a practice that courts have not yet blessed.
Section numbers refer to the bill text the analysis read — linked under Primary records below.
Who it affects
The bill creates transparency and accountability for AI training practices, potentially protecting creators' rights and enabling public scrutiny of AI development. However, the enforcement mechanism relies on copyright owners to sue, the $2.5 million annual cap may be insufficient deterrent for large AI firms, and the notice requirement does not grant creators compensation or consent rights—only disclosure and the ability to litigate.
Who stands to gain
- Copyright owners (ability to sue for non-disclosure)
- Law firms specializing in copyright litigation
- The Copyright Office (penalties fund operating costs)
Named in the bill
Register of Copyrights, Copyright Office, Generative AI model developers, Copyright owners, AMD, HPE, CTSH, CSCO, VRSN
Where it stands
1 cosponsor: 1 Republicans.
- Feb 10, 2026 — Introduced · Congress.gov: “Introduced in Senate”
- Feb 10, 2026 — Referred to Senate Committee on the Judiciary · Congress.gov: “Read twice and referred to the Committee on the Judiciary”
Dates and quoted wording are Congress.gov's action record; the timeline shows status changes, not every procedural step.
Money around this bill
6 lobbying clients named this bill on 7 disclosure filings across 1 quarter, Jun 2026 to Jun 2026. Those filings disclosed $4,397,139 in lobbying spend. A filing names 15 bills on average, so that figure is what each filing reported, not a share belonging to this bill.
More lobbying clients named this bill than 82% of bills with at least one filing.
Adam Schiff, the sponsor, reported $54,400 in PAC receipts in the 2026 cycle.
- Recording Industry Association of America — $2,206,361 on 1 filing
- Universal Music Group — $1,460,000 on 2 filings
- News Media Alliance — $475,000 on 1 filing
- International Alliance of Theatrical Stage Employees — $135,000 on 1 filing
- Recording Academy — $120,778 on 1 filing
Lobbying Disclosure Act filings through Jul 21, 2026. A filing shows who paid to lobby on a bill it names, not what changed.
How this was measured
Analysis — Quorum's AI read the bill text published by Congress.gov (4,405 characters) on Aug 17, 2026. Section numbers in the findings refer to that text, linked below; transparency and hidden-provision scores are compared against the median of 16,222 analysed bills.
Status and sponsors — Congress.gov's bill record — actions, committee referrals and cosponsors — loaded nightly. The timeline shows status changes, not every procedural action.
Money — Senate Lobbying Disclosure Act filings whose specific-issue field names this bill for quarters ending Jun 2026 to Jun 2026. A filing's amount is reported whole beside the median number of bills a filing names; it is never divided across them. PAC receipts are FEC-reported contributions to the sponsor's candidate committee in the 2026 cycle.
As of — lobbying records through Jul 21, 2026 · page rendered 2026-10-20.
“AI companies must disclose copyrighted training data—but can still use it” QuorumCivic. https://share.quorumcivic.app/bill/119/s3813 Report an error
The week in bills
One email, Sunday: what Congress passed, what it changed, and who paid.
One link arrives to confirm the address. It is used for this email and nothing else, and every issue carries an unsubscribe link.