NY, US · AI law tracker
A6578 — NY, US
A6578 is an AI governance legislation from NY, currently introduced. The bill introduces a mandate for generative AI model developers to disclose training data summaries [2]. AIGI tracks 3 primary-source updates on this bill; the most recent was published on 2025-03-06.
Status & timeline
- Regulatory stage
- introduced
- Bill status
- Assembly Floor Calendar
- Authority / governing body
- New York State Senate
- Chamber
- Assembly
- Document type
- legislation
Next deadline: No fixed deadline — bill is currently under legislative review.
Subscriber only
Full obligation matrix
| Actor | Obligation | Deadline | Source |
|---|---|---|---|
| provider | Developers of generative artificial intelligence models or services must post on their website information regarding the data used to train the model or service. | null | — |
| provider | Developers must include a high-level summary of the datasets used in the development of such generative AI system or service. | null | — |
Subscriber only
Enforcement risk score
Hypothetical or aspirational; no enforcement signal yet.
Subscriber only
Role-based compliance checklist
- cto Monitor the legislative progress of NY A6578.
- compliance_officer Assess current data governance and transparency practices regarding AI model training data.
- product_manager Begin strategizing how to compile and present a high-level summary of generative AI training datasets.
Subscriber only
Vendor impact assessment
- Vendor risk class
- medium
- Procurement categories
- foundation_models
Vendors providing generative AI models or services will likely face direct obligations to disclose training data information, impacting their product development and communication strategies. Buyers of these services should assess vendor readiness for such transparency requirements.
Sample vendor questions
- How do you plan to comply with potential training data transparency requirements, such as those proposed in NY A6578?
- Can you provide a high-level summary of the datasets used to train your generative AI models?
- What processes do you have in place for documenting and updating information about your training data sources?
- How would you support us in meeting our own transparency obligations regarding your AI models?
Intelligence briefs (3)
New York A6578: Establishes AI Training Data Transparency Act
The bill introduces a mandate for generative AI model developers to disclose training data summaries [2].
This proposal signals a growing legislative focus on transparency for generative AI foundation models.
Deadline: No fixed deadline — bill is currently under legislative review.
Primary source →NY A6578: Generative AI Training Data Transparency Bill Passes Assembly
The bill introduces a requirement for generative AI developers to post training data summaries online [1].
This signals an emerging legislative focus on transparency for foundational AI models, impacting developer obligations and public disclosure strategies.
Deadline: No fixed deadline — effective upon enactment
Primary source →New York A6578: Proposed AI Training Data Transparency Act
A New York Assembly bill proposes that generative AI developers disclose high-level summaries of training datasets on their websites [1].
This bill introduces a new transparency obligation for generative AI model developers, signaling an emerging regulatory focus on data provenance and intellectual property.
Deadline: No fixed deadline — bill is currently in committee.
Primary source →Frequently asked questions
- What is A6578?
- New York Assembly Bill A6578 (2025-2026) seeks to establish the Artificial Intelligence Training Data Transparency Act [1]. This proposed legislation mandates that developers of generative artificial intelligence models or services publicly disclose information about the data used for training their systems [2], specifically requiring a high-level summary of the datasets on their website. Primary source →
- Why does A6578 matter?
- This proposal signals a growing legislative focus on transparency for generative AI foundation models. Primary source →
- Who does A6578 affect?
- This proposed legislation primarily affects developers of generative artificial intelligence models or services operating or offering services within New York State. Organizations involved in the training and deployment of large language models, image generation models, or other generative AI systems would fall within scope. Considerations for these entities include their public disclosure practices and compliance with new transparency mandates. Primary source →
- What are the key dates for A6578?
- No fixed deadline — bill is currently under legislative review. Primary source →
- What is the current status of A6578?
- As of the last published update, A6578 is at the "introduced" stage, with bill status "Assembly Floor Calendar". Primary source →
- Where can I find the primary source for A6578?
- The primary source for the most recent update is at https://legislation.nysenate.gov/bills/2025/A6578. AIGI publishes the full citation chain plus every approved brief on this bill. Primary source →
Related
Stay informed