New York A 6578 (AI Training Data Transparency Act)

• Authority: New York
• Parent authority: State governments
• Jurisdiction: United States
• Collections: U.S. state and local documents
• Most recent activity: Proposed
• Most recent activity date: 2026-02-24

Risk subdomain coverage

Good coverage — Comprehensive governance measures or mitigations with clear procedures relating to the risk subdomain are explicitly described.

• Lack of transparency or interpretability

Minimal coverage — Brief mention but no specific mitigations, recommendations or governance measures relating to the risk subdomain are described.

• Compromise of privacy
• Economic and cultural devaluation of human effort

All other risk subdomains had no coverage in this document.

Risk Coverage Summary

The document primarily addresses transparency risks (7.4) with good coverage, requiring comprehensive disclosure of AI training data. It has minimal coverage of privacy risks (2.1) and intellectual property/creative work devaluation risks (6.3) through required disclosures about personal information and copyrighted content in training datasets.

Sector coverage

Minimal coverage — Brief mention but no specific mitigations, recommendations or governance measures relating to the sector are described.

• Trade, Transportation & Utilities
• Information
• Scientific Research & Development Services
• Public Administration (excl. National Security)
• National Security

All other sectors had no coverage in this document.

Sectors Summary

This is a cross-sectoral legislative act that applies to developers of generative AI models across all industries making AI available to New Yorkers. It explicitly exempts only aircraft operation (Trade, Transportation and Utilities) and national security applications (National Security sector). The law governs AI developers regardless of their industry sector, making it broadly applicable rather than sector-specific.

Legislative Status

Hard Law

Legislative Status Summary

This is a binding legislative act that amends New York's general business law with mandatory requirements for AI developers. It uses mandatory language ('shall'), establishes specific obligations with deadlines (January 1, 2027), and creates legally enforceable requirements as part of state law.

Actors

Proposer

• Entity Types: AI_Governance_Actor
• Summary: The document is a legislative act proposed by the New York State Legislature, as indicated by the enacting clause.

Target

• Entity Types: AI_Developer
• Summary: The act explicitly targets developers who design, code, produce, or substantially modify generative AI models or services for public use. This includes persons, partnerships, state or local government agencies, or corporations.

Enforcer

• Entity Types: AI_Governance_Actor
• Summary: The enforcement body is not explicitly named in the document. As an amendment to New York's general business law, enforcement would typically be handled by state authorities such as the Attorney General's office or a designated regulatory agency.

Monitor

• Entity Types: AI_Governance_Actor
• Summary: The document does not explicitly designate a monitoring body. The compliance mechanism relies on public website disclosure, which could be monitored by various state agencies or through public complaints.

AI Lifecycle Stages

Lifecycle stages covered by the documents, as defined in the OECD Framework for the Classification of AI Systems

• Collect and Process Data
• Build and Use Model
• Deploy

AI Lifecycle Summary

The document primarily covers the data collection and processing stage, with significant focus on the model building and training phase. It also applies requirements at the deployment stage by requiring documentation before models are made publicly available. The act does not address planning, verification, or ongoing monitoring stages.

Technical Scope

Terms relating to technical scope that are mentioned or implicitly referenced in the document

• AI Models
• Generative AI

Technical Scope Summary

The document explicitly defines and focuses on generative artificial intelligence models and AI models more broadly. It does not mention frontier AI, general purpose AI, foundation models, task-specific AI, predictive AI, compute thresholds, or open-weight models. AI systems are not explicitly defined as distinct from AI models.