New York Artificial Intelligence Training Data Transparency Act (A.6578-B)

Properties

jurisdiction: US, New York
level: state
domain: transparency
status: proposed
scope: Adds a new Article 44-C to the General Business Law requiring developers of generative AI models or services made publicly available to New Yorkers (released or substantially modified on or after January 1, 2022) to post training-data documentation on their website.
obligations: By January 1, 2027, and before each subsequent public release or substantial modification, a developer must post on its website documentation about the data used to train the model or service, including a high-level summary of datasets used: their sources or owners, how they further the model's intended purpose, number of data points (ranges allowed), types of data points, whether protected by copyright/trademark/patent or entirely public domain, whether purchased or licensed, whether they include personal or aggregate consumer information, any cleaning or processing performed, the data collection time period, and whether synthetic data generation was used. Exempts models solely for aircraft operation or made available only to a federal entity for national security, military, or defense purposes.
appliesTo: Developers that design, code, produce, or substantially modify a generative AI model or service made publicly available to New Yorkers
officialUrl: https://www.nysenate.gov/legislation/bills/2025/A6578
notes: Sponsored by M. of A. Bores (companion Senate bill S.6955-A). Passed the Assembly on May 5, 2026 and passed the Senate 54-6 on June 4, 2026 (independently confirmed against the bill's own recorded vote tally on nysenate.gov). Per the bill's own actions log and status tracker on nysenate.gov, it has passed both chambers but had not yet been delivered to Governor Hochul as of this check (August 2026); the tracker's Delivered to Governor checkpoint remains unchecked and no delivery action appears in the log, so the governor's decision clock has not started. New York's customary end-of-session practice gives the governor until December 31 of the year of delivery to sign or veto, but that is not yet an operative deadline here since delivery has not occurred; treat this as a general convention, not a confirmed date for this bill. Bill text (Section 2) states the act 'shall take effect immediately' upon signing (independently confirmed), with the first disclosure deadline of January 1, 2027 built into the substantive Section 1432. No dedicated penalty or enforcement section was found in the bill text reviewed. Distinct from already-tracked ca-ab2013 (California's near-identical training-data-transparency law): this is New York's own, separate state statute, still pending gubernatorial action, explicitly flagged as an untracked lead in this dataset's own ny-fair-news-act entry notes.

Properties

jurisdiction

US, New York

level

state

domain

transparency

status

proposed

scope

Adds a new Article 44-C to the General Business Law requiring developers of generative AI models or services made publicly available to New Yorkers (released or substantially modified on or after January 1, 2022) to post training-data documentation on their website.

obligations

By January 1, 2027, and before each subsequent public release or substantial modification, a developer must post on its website documentation about the data used to train the model or service, including a high-level summary of datasets used: their sources or owners, how they further the model's intended purpose, number of data points (ranges allowed), types of data points, whether protected by copyright/trademark/patent or entirely public domain, whether purchased or licensed, whether they include personal or aggregate consumer information, any cleaning or processing performed, the data collection time period, and whether synthetic data generation was used. Exempts models solely for aircraft operation or made available only to a federal entity for national security, military, or defense purposes.

appliesTo

Developers that design, code, produce, or substantially modify a generative AI model or service made publicly available to New Yorkers

officialUrl

https://www.nysenate.gov/legislation/bills/2025/A6578

notes

Sponsored by M. of A. Bores (companion Senate bill S.6955-A). Passed the Assembly on May 5, 2026 and passed the Senate 54-6 on June 4, 2026 (independently confirmed against the bill's own recorded vote tally on nysenate.gov). Per the bill's own actions log and status tracker on nysenate.gov, it has passed both chambers but had not yet been delivered to Governor Hochul as of this check (August 2026); the tracker's Delivered to Governor checkpoint remains unchecked and no delivery action appears in the log, so the governor's decision clock has not started. New York's customary end-of-session practice gives the governor until December 31 of the year of delivery to sign or veto, but that is not yet an operative deadline here since delivery has not occurred; treat this as a general convention, not a confirmed date for this bill. Bill text (Section 2) states the act 'shall take effect immediately' upon signing (independently confirmed), with the first disclosure deadline of January 1, 2027 built into the substantive Section 1432. No dedicated penalty or enforcement section was found in the bill text reviewed. Distinct from already-tracked ca-ab2013 (California's near-identical training-data-transparency law): this is New York's own, separate state statute, still pending gubernatorial action, explicitly flagged as an untracked lead in this dataset's own ny-fair-news-act entry notes.