New York Opens Registration for Frontier AI Developers Under RAISE Act
New York opened a registration portal on Tuesday, 10 November 2026, for developers of large frontier AI models under the RAISE Act. According to the governor's office, large frontier developers must register through the RAISE Portal by 1 January 2027. The law requires them to publish safety frameworks, report critical safety incidents within 72 hours and issue public transparency reports. The Office of Digital Innovation, Governance, Integrity, and Trust (DIGIT) will run the law. For most businesses that automate work, the direct duties fall on the developers that build the models, not on the companies that buy access to them. The practical effect is more public documentation from the largest AI vendors, and a new route for checking how those vendors handle safety.
In short
- New York's RAISE Act registration portal opened on 10 November 2026, with a deadline of 1 January 2027 for large frontier developers.
- Covered developers must publish safety frameworks, report critical safety incidents within 72 hours and issue transparency reports.
- Penalties can reach $1 million for a first violation and $3 million for later violations, according to TokenPost.
- The direct duties fall on model developers, so businesses should focus on vendor documentation and their own approval steps.
- Watch DIGIT's rulemaking and its first annual incident report.
What happened
On Tuesday, 10 November 2026, the administration of Governor Kathy Hochul opened a registration portal for the RAISE Act, New York's AI safety and transparency law. According to TechCrunch, large frontier developers such as OpenAI and Anthropic must register by 1 January 2027. The governor's office described the portal as open to large frontier AI developers and said registration is required by that date. The same day, the state launched the New York Center for AI Responsibility and Research at Binghamton University, which will independently test advanced AI systems and help review developers' safety frameworks.
The law applies to companies with more than $500 million in annual revenue, according to TechCrunch. Those companies must publish safety frameworks on their websites, report critical safety incidents within 72 hours and issue public model transparency reports. TokenPost reports that penalties can reach $1 million for a first violation and $3 million for later violations. The state can also bring enforcement actions against companies that fail to register, according to Mezha, although a spokesperson declined to give details.
Why it matters for businesses that automate work
Most businesses do not build frontier models. They buy access to them, through chat assistants, document tools, customer messaging bots and back-office software that calls a model in the background. The RAISE Act does not set the rules for how a business runs its own workflows. It does change what the large model makers must disclose, and that information can shape which vendors a business chooses and how it writes its own AI policy.
There is also a timing point. The registration deadline falls in late 2026, but the reporting duties start in January 2027. The governor's office says the law's duties start with the new year. Businesses that plan a vendor review in early 2027 will have published frameworks and transparency reports to compare, if the vendors meet the deadline.
What changes in practice
For a business automating work, the change is mostly in what you can read and what you should ask. Start with a short list of every AI tool that touches your records. For each one, note the vendor, the model name and the tasks it performs, such as drafting follow-up emails, matching invoices or sorting support tickets.
Then check whether the vendor publishes a safety framework and a transparency report once the law is in force. Ask for the report and keep a copy with the date you received it. This gives your finance or operations lead a dated record when a vendor changes its model.
Build your workflows so a person checks the output where money, pay or customer obligations are involved. Payment matching, payroll entries and billing should keep a human approval step. A safety report from a vendor does not check the accuracy of your own data, and it does not replace your own review.
| Question for your team | Why it matters |
|---|---|
| Which AI tools touch invoices, payroll or customer records? | These are the places where an error costs money or breaks a promise. |
| Has the vendor published a safety framework and transparency report? | Under the RAISE Act these documents become public for covered developers from January 2027. |
| What happens to our workflow if the model changes? | Incident reports can lead to model changes, so you need a fallback step. |
| Who in our business receives vendor notices? | A notice that reaches nobody is no notice at all. |
Limits of this change
The RAISE Act is a state law. It targets large frontier developers, so a small company that only uses a model through a standard subscription does not get new duties from it. Your own obligations under other laws, including employment, privacy and sector rules, stay exactly as they were. The law also does not guarantee that any model is safe. It sets reporting and documentation duties, which the state will review.
What to watch next
Three dates matter. The first is 1 January 2027, when registration closes and the reporting duties start, according to the governor's office. The second is the first annual report DIGIT will publish on the incidents it receives, which the public can also report to. The third is any rulemaking by the Department of Financial Services. The law gives the office broad authority to adopt rules, and Davis Polk notes that the agency may allow developers to use federal reporting standards in place of the state report. Businesses should watch those rules, because they decide how the 72-hour duty works in practice.
How AiStaffo would automate this
AiStaffo would connect the AI tools a business already uses to a single log that records the vendor, model name, task and date of each change. The workflow runs automatically: it collects vendor notices and published documents, files them, and flags any change to a tool that touches invoices, payments or customer records. A person still reviews those flags, decides whether a workflow needs a fallback, and signs off on any change to payment or payroll steps. Results depend on the reader's own tools and volumes, which the free audit measures. Book a free automation audit.
Questions people ask
What is the RAISE Act?
When do companies have to register under the RAISE Act?
Does the RAISE Act apply to my business?
What is DIGIT?
How does the RAISE Act compare with California's SB 53?
What should a business do now?
Book a free automation audit
Thirty minutes. We look at one process you run every week and tell you exactly what an AI worker would take off your desk, and what it would not.
