The opening of the registration portal in New York establishes a binding regulatory mechanism to collect safety frameworks and serious incident reports from frontier model developers instead of relying solely on voluntary press releases.
Opening of the registration infrastructure and scope of companies covered by the RAISE Act
Requirements for publishing safety frameworks and the 72-hour reporting deadline
Regulatory debate between state laws and U.S. federal rules
The role of academic evaluation at Binghamton University and upcoming milestones
Frequently asked questions
Opening of the registration infrastructure and scope of companies covered by the RAISE Act
On November 10, 2026, the State of New York opened the official registration infrastructure associated with the «RAISE» Act for frontier AI safety. As TechCrunch reported in its coverage of the rollout, developers meeting the law’s criteria face new obligations regarding documentation and incident reporting.
The legislation specifically targets major developers and does not include every small business using an AI assistant in its daily operations; the law’s scope depends on precisely defined revenue thresholds and whether an organization is developing the covered category of advanced models.
Requirements for publishing safety frameworks and the 72-hour reporting deadline
Legal requirements include publishing approved safety frameworks, disclosing specific serious incidents, and providing state authorities with the information necessary to exercise oversight. Press coverage indicates that severe incidents are expected to be reported within 72 hours when relevant conditions are met, though organizations must consult the final legislative text and implementation guidelines to determine exactly what must be reported, as not every ordinary software bug constitutes a mandatory reportable catastrophic safety incident.
The law focuses on pre-deployment and post-deployment accountability; a written safety policy outlines what a company intends to evaluate, but it does not prove by itself that a model has undergone sufficient testing. This necessitates enabling regulators and independent researchers to review clear evidence of tests, known limitations, and mechanisms for handling severe failures, while accounting for varying risks across model architectures and deployment environments.
Regulatory debate between state laws and U.S. federal rules
New York’s law intersects with the escalating debate in the United States over state-level regulation versus unified federal oversight; a developer selling products nationwide may face divergent obligations across different states.
Industry companies argue that fragmented rules increase compliance complexity and costs, while proponents of state-level action maintain that additional oversight is necessary as long as federal safeguards remain limited. These arguments remain policy positions rather than technical metrics.
The role of academic evaluation at Binghamton University and upcoming milestones
Tied to this path is an academic evaluation effort at «Binghamton University» aimed at supporting the analysis of advanced systems; public research can make performance claims more testable, though factors such as independence, funding, methodologies, and model access remain crucial to evaluation quality. The portal’s opening on November 10 marks the new operational step, while the legislative texts and associated funding announcements date back to earlier periods.
The importance of this step lies in establishing a practical mechanism for gathering information from frontier model developers instead of relying exclusively on voluntary press releases. Upcoming milestones will include the completion of registrations, the publication of safety frameworks, incident reports, and any enforcement actions.
On the legislative and regulatory front in the United States, the activation of the registration portal in New York represents a tangible transition from theoretical debate regarding frontier model risks to actual administrative enforcement. The coming months will reveal the extent to which mandatory disclosure mechanisms and the mandated 72-hour deadline for reporting severe incidents—integrated with academic evaluations at Binghamton University—can achieve genuine, measurable, and accountable transparency without disrupting the innovation ecosystem. As legal and technical circles await how major labs will interact with New York’s new registration requirements, the practical experience of this law will form an important legislative reference for other U.S. states considering enacting similar controls to oversee frontier systems and close federal regulatory gaps. Independent researchers and legislative bodies will also monitor the clarity of periodic reports published via the official portal and their ability to uncover actual risks of advanced systems before they occur.
Frequently asked questions
Question: Does New York’s RAISE Act apply to all companies using artificial intelligence?
Answer: No; the law targets major frontier model developers only, based on specific revenue thresholds and the development of certain categories of massive models, and does not include small businesses using tools.
Question: What are the primary obligations imposed by the RAISE Act on covered developers?
Answer: It requires them to officially register, publish safety frameworks, provide oversight information to the state, and report specified severe incidents within a 72-hour window under legal conditions.
Question: Is publishing a written safety policy sufficient proof of a model’s safety?
Answer: No; a written policy only outlines evaluation intentions, and regulators and independent researchers require clear test evidence, details on known limitations, and how failures are addressed.
Question: What is the role of Binghamton University in New York’s new oversight ecosystem?
Answer: The university is participating through a supporting academic evaluation effort to analyze advanced artificial intelligence systems and make performance and safety claims more verifiable through scientific examination.