The New York RAISE Act (Responsible AI Safety and Education Act) is a state law establishing transparency and safety requirements for developers of frontier AI models. It was authored by Sen. Gounardes and introduced in the New York State Legislature on January 8, 2026, where it was referred to the Committee on Rules. The bill is closely modeled on California SB 53. The RAISE Act was enacted: Governor Kathy Hochul signed the underlying bill on December 19, 2025, then signed a negotiated chapter amendment (carried as S8828) on March 27, 2026; the final law takes effect January 1, 2027 (Source: https://www.wiley.law/alert-New-York-Finalizes-RAISE-Act-for-Frontier-AI-Models-Law-Takes-Effect-January-1-2027).
Status and legislative history
The New York State Legislature passed the RAISE Act in June 2025, sponsored by Assemblymember Alex Bores and Senator Andrew Gounardes (Source: https://fpf.org/blog/the-raise-act-vs-sb-53-a-tale-of-two-frontier-ai-laws/). Governor Hochul signed the bill on December 19, 2025, making New York the second state, after California, to enact a statute specifically focused on frontier AI safety and transparency (Source: https://fpf.org/blog/the-raise-act-vs-sb-53-a-tale-of-two-frontier-ai-laws/). At signing, Hochul described the law as building on California's framework and "creating a unified benchmark among the country's leading tech states as the federal government lags behind" (Source: https://www.governor.ny.gov/news/governor-hochul-signs-nation-leading-legislation-require-ai-frameworks-ai-frontier-models).
The December signing was paired with a negotiated chapter amendment that revised the enacted text. The chapter amendment was introduced on January 6, 2026, passed the second chamber of the legislature on March 11, 2026, and was signed by Governor Hochul on March 27, 2026, as the final version of the law (Source: https://www.wiley.law/alert-New-York-Finalizes-RAISE-Act-for-Frontier-AI-Models-Law-Takes-Effect-January-1-2027). The chapter amendment narrowed the differences between the RAISE Act and California's SB 53, including by adopting the same computational-operations and revenue thresholds (Source: https://www.wiley.law/alert-New-York-Finalizes-RAISE-Act-for-Frontier-AI-Models-Law-Takes-Effect-January-1-2027). The final law takes effect January 1, 2027, after SB 53, which became operative January 1, 2026 (Source: https://fpf.org/blog/the-raise-act-vs-sb-53-a-tale-of-two-frontier-ai-laws/).
Scope and definitions
The law applies to frontier models "developed, deployed, or operating" in New York and reaches only models developed or operated in whole or in part in the state, a territorial limitation not present in SB 53 (Source: https://www.wiley.law/alert-New-York-Finalizes-RAISE-Act-for-Frontier-AI-Models-Law-Takes-Effect-January-1-2027; Source: https://fpf.org/blog/the-raise-act-vs-sb-53-a-tale-of-two-frontier-ai-laws/). The Future of Privacy Forum notes that the territorial limitation may make the RAISE Act more likely than SB 53 to survive a Dormant Commerce Clause challenge (Source: https://fpf.org/blog/the-raise-act-vs-sb-53-a-tale-of-two-frontier-ai-laws/). The law also includes explicit carveouts for universities engaged in research (Source: https://fpf.org/blog/the-raise-act-vs-sb-53-a-tale-of-two-frontier-ai-laws/).
The statute defines its covered actors and technologies as follows (Source: https://fpf.org/blog/the-raise-act-vs-sb-53-a-tale-of-two-frontier-ai-laws/; Source: https://www.wiley.law/alert-New-York-Finalizes-RAISE-Act-for-Frontier-AI-Models-Law-Takes-Effect-January-1-2027):
- Foundation model — an AI model trained on a broad data set, designed for generality of output, and adaptable to a range of tasks.
- Frontier model — a foundation model trained using computing power greater than 10^26 integer or floating-point operations.
- Frontier developer — a person who trained, or initiated the training of, a frontier model.
- Large frontier developer — a frontier developer with annual gross revenue exceeding $500 million, the category targeted by the law's most extensive obligations.
These compute and revenue thresholds match those used in California's SB 53; the chapter amendment added them in place of the original bill's compute-spending threshold (Source: https://www.wiley.law/alert-New-York-Finalizes-RAISE-Act-for-Frontier-AI-Models-Law-Takes-Effect-January-1-2027).
The law is organized around preventing "catastrophic risk," defined as a foreseeable and material risk that a frontier developer's development, storage, use, or deployment of a frontier model will materially contribute to the death of or serious injury to more than 50 people, or more than $1 billion in damage to or loss of property, arising from a single incident in which the model does any of the following (Source: https://www.wiley.law/alert-New-York-Finalizes-RAISE-Act-for-Frontier-AI-Models-Law-Takes-Effect-January-1-2027; Source: https://fpf.org/blog/the-raise-act-vs-sb-53-a-tale-of-two-frontier-ai-laws/):
- Provides expert-level assistance in creating or releasing a chemical, biological, radiological, or nuclear weapon;
- Engages, without meaningful human oversight, in conduct that is a cyberattack or that, if committed by a human, would constitute murder, assault, or theft; or
- Evades the control of its developer or user in a manner that demonstrates materially increased catastrophic risk.
The definition excludes risk from the release of otherwise publicly available information, lawful activity of the federal government, and harm to which the model did not materially contribute (Source: https://www.wiley.law/alert-New-York-Finalizes-RAISE-Act-for-Frontier-AI-Models-Law-Takes-Effect-January-1-2027). A "critical safety incident" includes unauthorized access to, modification of, or exfiltration of model weights resulting in death or bodily injury; harm resulting from the materialization of a catastrophic risk; loss of control of the model causing death or bodily injury; and use of deceptive techniques by the model against its developer in a manner demonstrating materially increased catastrophic risk (Source: https://www.wiley.law/alert-New-York-Finalizes-RAISE-Act-for-Frontier-AI-Models-Law-Takes-Effect-January-1-2027).
Key provisions
The bill requires frontier AI developers to:
- Publish safety frameworks with standardized, mandatory disclosures
- Conduct pre-release safety evaluations
- Report critical safety incidents to government
- Take due care proportional to the scale of foreseeable risks
The legislative findings state that "not all developers are providing information that is consistent and sufficient to ensure necessary transparency and protection of the public," which the bill offers as the justification for mandatory rather than voluntary disclosure.
The enacted obligations are structured as follows, with certain duties applying to all frontier developers and additional duties reserved for large frontier developers (Source: https://fpf.org/blog/the-raise-act-vs-sb-53-a-tale-of-two-frontier-ai-laws/; Source: https://www.wiley.law/alert-New-York-Finalizes-RAISE-Act-for-Frontier-AI-Models-Law-Takes-Effect-January-1-2027):
- Frontier AI Framework — Large frontier developers must publish on their website a framework describing how they incorporate national standards, set thresholds and mitigations for catastrophic risk, review and assess those mitigations, use third parties to conduct assessments, update the framework, secure unreleased model weights through cybersecurity practices, maintain internal governance, identify and respond to critical safety incidents, and manage catastrophic risk arising from internal use of the model. The framework must be updated at least annually or sooner after a material modification. Limited redactions are permitted to protect trade secrets, cybersecurity, public safety, or national security (Source: https://www.wiley.law/alert-New-York-Finalizes-RAISE-Act-for-Frontier-AI-Models-Law-Takes-Effect-January-1-2027; Source: https://fpf.org/blog/the-raise-act-vs-sb-53-a-tale-of-two-frontier-ai-laws/).
- Transparency report — Before deploying a new frontier model or a modified version of an existing model, all frontier developers (not only large developers) must publish a transparency report giving developer contact information, the model's release date, supported languages and output modalities, intended use, restrictions or conditions on use, and summaries of catastrophic-risk assessments and the role of any third-party evaluators. Publishing the information in a system card or model card is deemed compliant; redaction is permitted on the same grounds as the framework, subject to justification and recordkeeping (Source: https://www.wiley.law/alert-New-York-Finalizes-RAISE-Act-for-Frontier-AI-Models-Law-Takes-Effect-January-1-2027).
- Safety incident reporting — Frontier developers must report critical safety incidents to the DFS oversight office within 72 hours of determining that one occurred or of learning facts sufficient to form a reasonable belief that one occurred. Where an incident poses an imminent risk of death or serious physical injury, disclosure to an appropriate law-enforcement or public-safety agency is required within 24 hours, and the developer must still report to the Office within 72 hours. The law also requires a mechanism for public reporting of incidents (Source: https://www.wiley.law/alert-New-York-Finalizes-RAISE-Act-for-Frontier-AI-Models-Law-Takes-Effect-January-1-2027; Source: https://fpf.org/blog/the-raise-act-vs-sb-53-a-tale-of-two-frontier-ai-laws/).
- Internal-use risk summaries — Large frontier developers must transmit to the Office a summary of any assessment of catastrophic risk resulting from internal use of their frontier models every three months, with written updates. The Office must allow confidential submission and limit access to authorized personnel (Source: https://www.wiley.law/alert-New-York-Finalizes-RAISE-Act-for-Frontier-AI-Models-Law-Takes-Effect-January-1-2027).
- Frontier developer disclosure statement — A large frontier developer may not operate a frontier model in New York without filing a disclosure statement with the Office, renewed at least every two years and upon a change in ownership or material change. The statement must identify the developer and its business names, the addresses of its principal place of business and each New York office, and all persons holding (or who held in the past five years, if privately held) a 5 percent or greater beneficial ownership interest. The Future of Privacy Forum describes this disclosure program, which requires information such as ownership structure, as a New York-specific addition not present in SB 53 (Source: https://www.wiley.law/alert-New-York-Finalizes-RAISE-Act-for-Frontier-AI-Models-Law-Takes-Effect-January-1-2027; Source: https://fpf.org/blog/the-raise-act-vs-sb-53-a-tale-of-two-frontier-ai-laws/).
Oversight office and rulemaking
The law creates a new oversight office within the New York State Department of Financial Services (DFS) to assess large frontier developers and receive the required reports and summaries (Source: https://www.governor.ny.gov/news/governor-hochul-signs-nation-leading-legislation-require-ai-frameworks-ai-frontier-models). The Office is granted rulemaking authority to implement the law, including authority to consider additional reporting or publication requirements and to designate federal regulations, laws, or guidance as methods of compliance with certain reporting obligations (Source: https://www.wiley.law/alert-New-York-Finalizes-RAISE-Act-for-Frontier-AI-Models-Law-Takes-Effect-January-1-2027). The Future of Privacy Forum contrasts this direct DFS rulemaking authority with SB 53, under which the California Department of Technology recommends definitional updates rather than issuing rules directly (Source: https://fpf.org/blog/the-raise-act-vs-sb-53-a-tale-of-two-frontier-ai-laws/).
The Office may share reports and internal-use risk summaries with other government agencies, including the New York Attorney General, subject to statutory factors. Beginning in January 2028, the Office must produce an annual report for the Governor and Legislature using anonymized and aggregated critical-safety-incident information, with recommendations on updates to the law, while withholding information that would compromise trade secrets, cybersecurity, public safety, or national security (Source: https://www.wiley.law/alert-New-York-Finalizes-RAISE-Act-for-Frontier-AI-Models-Law-Takes-Effect-January-1-2027).
Enforcement and penalties
The Attorney General enforces the law through civil actions, with penalties of up to $1 million for a first violation and up to $3 million for subsequent violations, scaled to the severity of the offense (Source: https://www.governor.ny.gov/news/governor-hochul-signs-nation-leading-legislation-require-ai-frameworks-ai-frontier-models; Source: https://www.wiley.law/alert-New-York-Finalizes-RAISE-Act-for-Frontier-AI-Models-Law-Takes-Effect-January-1-2027). Violations include failure to file a required document, false or misleading statements about a developer's Frontier AI Framework, failure to report an incident, and failure to comply with a developer's own framework. Separately, the Office may impose civil penalties of $1,000 per day, after notice and a hearing, for failure to file a disclosure statement or correct false information (Source: https://www.wiley.law/alert-New-York-Finalizes-RAISE-Act-for-Frontier-AI-Models-Law-Takes-Effect-January-1-2027). The statute creates no private right of action, and it permits a large frontier developer to assert that alleged harm was caused by another person, entity, or contributing factor, a clarification the Future of Privacy Forum notes is absent from SB 53 (Source: https://www.wiley.law/alert-New-York-Finalizes-RAISE-Act-for-Frontier-AI-Models-Law-Takes-Effect-January-1-2027; Source: https://fpf.org/blog/the-raise-act-vs-sb-53-a-tale-of-two-frontier-ai-laws/).
Relation to other legislation
The bill is positioned as the East Coast counterpart to California's SB 53, the two together creating concurrent state-level requirements on frontier AI developers. Before the text was available, the bill's transparency-first approach had been described through Dario Amodei's characterizations; the text confirms that approach, which Anthropic supported and helped craft. The RAISE Act is an instance of techno-federalism, in which states lead AI regulation in the absence of comprehensive federal action.
The Future of Privacy Forum's comparison finds the two statutes share largely identical scope and core requirements after the chapter amendment, while several distinctions remain: SB 53 includes employee whistleblower protections that the RAISE Act lacks; the RAISE Act adds the frontier developer disclosure program; SB 53 allows 15 days for safety-incident reporting against the RAISE Act's 72-hour window; and the RAISE Act authorizes higher penalties ($1 million / $3 million versus SB 53's $1 million per violation) (Source: https://fpf.org/blog/the-raise-act-vs-sb-53-a-tale-of-two-frontier-ai-laws/). Wiley Rein similarly characterizes the 72-hour reporting requirement as a marked departure from the California Transparency in Frontier Artificial Intelligence Act (TFAIA, the operative name for SB 53), which uses a 15-day timeline, resulting in a distinct New York standard (Source: https://www.wiley.law/alert-New-York-Finalizes-RAISE-Act-for-Frontier-AI-Models-Law-Takes-Effect-January-1-2027).
Before the chapter amendment, the RAISE Act diverged more sharply from SB 53. The version passed by the legislature applied to "large developers" defined by annual compute spending above $100 million rather than the frontier-developer and large-frontier-developer distinction; framed its focus around "critical harm" with heightened liability qualifiers rather than the shared "catastrophic risk" standard; lacked the transparency-report obligation and the disclosure program; required a less prescriptive "safety and security protocol" in place of the Frontier AI Framework; included a deployment prohibition barring models posing an unreasonable risk of critical harm, which SB 53 does not contain; and set penalties up to $10 million for a first violation and $30 million for subsequent violations. The chapter amendment removed the deployment prohibition, replaced the liability qualifiers with the catastrophic-risk standard, and scaled the penalties down to the enacted $1 million / $3 million figures (Source: https://fpf.org/blog/the-raise-act-vs-sb-53-a-tale-of-two-frontier-ai-laws/).
Both SB 53 and the RAISE Act include severability clauses. The Future of Privacy Forum observes that these clauses may become more consequential amid federal-state tension over AI policy, citing a December 2025 White House executive order that seeks to apply federal pressure against state AI laws deemed excessive, including through an AI Litigation Task Force and funding restrictions tied to state enforcement (Source: https://fpf.org/blog/the-raise-act-vs-sb-53-a-tale-of-two-frontier-ai-laws/). Other states, including Michigan, have introduced similar bills (Source: https://fpf.org/blog/the-raise-act-vs-sb-53-a-tale-of-two-frontier-ai-laws/).
Reactions
At signing, sponsors framed the law as exceeding California's. Assemblymember Bores said New York "now has the strongest AI transparency law in the country" and that the bill "moves beyond California's SB53 in significant ways" (Source: https://www.governor.ny.gov/news/governor-hochul-signs-nation-leading-legislation-require-ai-frameworks-ai-frontier-models). Senator Gounardes said the law "lays the groundwork for a world where AI innovation makes life better instead of putting it at risk" (Source: https://www.governor.ny.gov/news/governor-hochul-signs-nation-leading-legislation-require-ai-frameworks-ai-frontier-models). DFS Acting Superintendent Kaitlin Asrow said the department "looks forward to supporting Governor Hochul's continued efforts to foster innovation and establish standards for the safe development of artificial intelligence models" (Source: https://www.governor.ny.gov/news/governor-hochul-signs-nation-leading-legislation-require-ai-frameworks-ai-frontier-models).
Sources
- Full text: S. 8828 (chapter amendment), New York State Legislature; underlying bill S6953/A6453 (introduced 2026-01-08 as filed in this vault)
- Governor Hochul signing announcement, December 19, 2025 (Source: https://www.governor.ny.gov/news/governor-hochul-signs-nation-leading-legislation-require-ai-frameworks-ai-frontier-models)
- Future of Privacy Forum, "The RAISE Act vs. SB 53: A Tale of Two Frontier AI Laws," January 8, 2026 (Source: https://fpf.org/blog/the-raise-act-vs-sb-53-a-tale-of-two-frontier-ai-laws/)
- Wiley Rein, "New York Finalizes RAISE Act for Frontier AI Models; Law Takes Effect January 1, 2027," April 3, 2026 (Source: https://www.wiley.law/alert-New-York-Finalizes-RAISE-Act-for-Frontier-AI-Models-Law-Takes-Effect-January-1-2027)