New York S8828 requires large frontier AI developers to publish a Frontier AI Framework covering catastrophic-risk assessment and incident response, to publish transparency reports before deploying new frontier models, and to report critical safety incidents within 72 hours. It was signed on 27 March 2026 as Chapter 96 and takes effect on 1 January 2027.
New York S8828, sponsored by Senator Andrew Gounardes, is summarised officially as implementing "transparency requirements for developers of AI models" and requiring the establishment of an office to oversee AI model developer transparency and reporting. The Governor signed it on 27 March 2026 as Chapter 96, and it takes effect on 1 January 2027.
The law applies to large frontier developers rather than to AI users or deployers generally. Those developers must publish a Frontier AI Framework describing how they assess catastrophic risk, what mitigations they apply, how they handle cybersecurity, and how they respond to critical safety incidents. The framework must be reviewed and updated annually, and a transparency report must be published before a new or substantially modified frontier model is deployed.
What makes the statute unusually concrete is that it defines the harm it is aimed at. Catastrophic risk is tied to thresholds — incidents causing 50 or more deaths, or at least $1 billion in property damage. That converts an otherwise open-ended safety obligation into something a developer can test a model against and a regulator can audit.
Reporting runs on a clock. Critical safety incidents must be reported to the Office within 72 hours, tightening to 24 hours where the risk is imminent, and developers must submit quarterly summaries of their internal catastrophic risk assessments.
For a company already publishing a safety framework voluntarily, much of this is documentation of existing practice. The change is that the framework becomes a filed, annually renewed obligation with a named regulator and a deadline attached to failure.
New York — what the law requires
The figures AI Law Tracker holds for New York — tracked jurisdiction-wide, not derived from this story. See every source on the New York law page →
Key requirement. NYC Local Law 144 requires annual independent bias audits for automated employment decision tools. Separately, the RAISE Act — signed December 2025, effective January 1, 2027 — imposes safety-protocol, incident-reporting, and oversight duties on large frontier-AI developers.
Jurisdiction record last verified Jul 4, 2026
Sources & citations
- The official bill summary, the sponsor, the Frontier AI Framework and annual review duty, pre-deployment transparency reports, the 50-deaths / $1 billion catastrophic-risk thresholds, the 72-hour and 24-hour incident reporting deadlines, quarterly risk-assessment summaries, the 27 March 2026 signing as Chapter 96, and the 1 January 2027 effective date. www.nysenate.gov
See the full AI law picture for New York
Statuses, deadlines, penalties, and primary sources — kept current on AI Law Tracker.
How this was made. This story is drafted by an automated pipeline from the primary sources cited above, then checked against those sources before it is published. Nothing is published as raw machine output, and no claim here is stronger than the source it links to. Full methodology · our own data & AI practices.