Verdict: The New York RAISE Act is the strictest of the three main US frontier-AI safety laws on incident speed. California's SB 53 was the template and Illinois leads on audit verification, but the first hard deadline is New York's. On November 10, 2026, New York opened its RAISE Portal, and large frontier developers must register through it by January 1, 2027. New York gives labs 72 hours to report a critical safety incident; California's SB 53 allows 15 days; Illinois adds independent audits. For Indian builders and enterprise buyers, the practical result is more public paperwork to read before signing an API or enterprise contract with a US lab.
TL;DR
- Large frontier developers — those with more than $500 million in annual revenue — must register through the RAISE Portal that New York's DIGIT office opened on November 10, 2026, by January 1, 2027.
- Critical safety incidents must be reported within 72 hours, the shortest window among the three state laws compared here, with penalties reported at $1 million for a first violation and $3 million for later ones (TechCrunch).
- New York also launched the state's own AI research center at Binghamton: 80-plus researchers who will independently test frontier systems.
- Indian teams gain public safety frameworks to use in vendor diligence; India itself has no equivalent registration regime yet.
What does the New York RAISE Act require from frontier labs?
The Responsible AI Safety and Education Act (S6953B/A6453B) was signed on December 19, 2025, with the heaviest duties on companies earning more than $500 million a year, enforced by the Office of Digital Innovation, Governance, Integrity, and Trust (DIGIT). TechCrunch names OpenAI and Anthropic as the two labs that must register.
According to the state's implementation announcement, the obligations from January 2027 are:
- Establish safety and transparency frameworks and publish them on the developer's website.
- Report critical safety incidents to the DIGIT office within 72 hours, and file quarterly assessments of catastrophic risk.
- Register with DIGIT, file a disclosure statement at least every other year starting January 1, 2027, and pay assessments.
Every frontier developer operating in New York must report critical safety incidents and publish model transparency reports; the published safety framework requirement falls on the large developers.
How does the New York RAISE Act compare with California SB 53 and Illinois SB 315?
California's SB 53 took effect in January 2026 and requires frontier labs to maintain and publish safety frameworks, protect whistleblowers, and report critical safety incidents within 15 days. Illinois SB 315 starts at the beginning of 2027 with similar requirements plus independent safety compliance audits (TechCrunch).
| New York RAISE Act | California SB 53 | Illinois SB 315 | |
|---|---|---|---|
| Large-developer threshold | More than $500 million annual revenue | Not stated in the cited reporting | Not stated in the cited reporting |
| Incident-report window | 72 hours | 15 days | Not specified in the cited reporting |
| Independent audits | Not among the listed obligations | Not among the reported requirements | Yes |
| Published safety framework | Yes, from January 2027 | Yes | Similar requirements |
| Status | Portal open, register by January 1, 2027 | In effect since January 2026 | Starts at the beginning of 2027 |
New York wins on early warning, Illinois on outside verification, California is the template both build on. Treat the empty cells as a prompt to read the statutes before relying on this table for legal purposes.
What happens on and after January 1, 2027?
January 1, 2027 is the first hard filing date any major US lab faces under a state frontier-AI law with a dedicated registration system. A covered lab's calendar, which doubles as a compliance-watch checklist:
- Register through the RAISE Portal by January 1, 2027.
- Publish the safety and transparency framework and the model transparency reports on the company website.
- File quarterly assessments of catastrophic risk with DIGIT, a disclosure statement at least every other year, and the assessments it requires.
- Keep the 72-hour incident clock running for any critical safety incident.
The January check for outside observers: whether OpenAI and Anthropic have published frameworks that match what the state describes. Companies with a public framework page are easy to verify. Companies without one are the story.
Who reviews the safety frameworks?
The New York Center for AI Responsibility and Research (NYCAIRR), launched at SUNY Binghamton University alongside the portal, supplies the technical expertise: 80-plus researchers across the state, interim co-directors Greg Morrisett of Cornell Tech and Jeannette Wing of Columbia University, running on Empire AI, the state's $500 million public-private compute initiative. It was first announced on January 21, 2026 with a $30 million philanthropic gift and a $25 million SUNY research investment (SUNY).
At the state's request, NYCAIRR can inform policy and will coordinate with DIGIT to review developers' published frameworks; the Department of Financial Services (DFS) leads implementation.
What are the penalties, and what is still unclear?
TechCrunch reports penalties of up to $1 million for one violation and $3 million for subsequent violations. The state can bring enforcement actions against companies that fail to register, but a spokesperson declined to give specifics on how that would work.
A published deadline is clear; the enforcement process is not, and it is not yet known how DIGIT will treat a framework that is published but thin. Governor Hochul framed the law as a response to federal inaction, saying the stakes are too high to take AI companies at their word.
What does the New York RAISE Act mean for Indian builders and enterprises?
India has no equivalent frontier-lab registration regime today. That does not make New York's rules irrelevant: OpenAI and Anthropic APIs sit under many Indian coding agents and copilots, and the safety frameworks they must publish become documents a procurement team or security reviewer can read before a deal closes.
Three practical uses:
- Add the published framework and transparency report to vendor due-diligence packs.
- Map your agent deployments to the vendor's stated risk tiers; the AI agent sandbox containment playbook covers the technical side.
- Expect frameworks to become negotiation artifacts like licensing terms already are: the Anthropic licensing deal with ANI in India shows these labs already negotiating India-specific legal ground.
Two companion reads: the best generative AI assistants of 2026 comparison and the OpenAI business hub for small business in India guide.
Frequently asked questions
Q: What is the New York RAISE Act?
A: The Responsible AI Safety and Education Act is a New York law, signed on December 19, 2025, that requires frontier AI developers to publish safety frameworks and report critical safety incidents, with the heaviest duties on companies earning more than $500 million a year.
Q: When is the New York RAISE Act registration deadline?
A: Large frontier developers must register through the RAISE Portal by January 1, 2027; the portal opened November 10, 2026.
Q: How fast must labs report incidents under the New York RAISE Act?
A: Within 72 hours of a critical safety incident, reported to the DIGIT office. California's SB 53 allows 15 days.
Q: What are the penalties for non-compliance?
A: TechCrunch reports up to $1 million for one violation and $3 million for subsequent ones.
Q: Does the New York RAISE Act apply to Indian companies?
A: It applies to frontier developers operating in New York. Indian companies building on OpenAI or Anthropic models are affected indirectly: the published frameworks become vendor-review documents.
Top comments (0)