New York lawmakers have introduced the Responsible AI Safety and Education (RAISE) Act, Assembly Bill A6453, which establishes new regulatory requirements for organizations developing or deploying certain high-risk artificial intelligence (AI) systems. The bill focuses on large-scale, high-cost AI models known as frontier models, and is intended to mitigate risks related to public safety, cybersecurity, and critical infrastructure.
If enacted, the law would apply 90 days after it is signed by the governor and affect organizations that develop or use qualifying AI systems in New York State.
Key Provisions of the RAISE Act
- Covered AI Systems. The bill applies to frontier models which are defined as advanced AI systems that meet high compute thresholds or cost over $100 million to train. It also includes certain systems derived from these models.
- Developer Requirements. “Large developers” would be required to:
- Create and maintain a written safety and security protocol.
- Publish a redacted version of that protocol and submit it to the Attorney General and Division of Homeland Security
- Implement safeguards to prevent unreasonable risk of critical harm, including those related to weapons, autonomous actions, and large-scale financial or physical damage.
- Report any safety incidents within 72 hours.
- Annual Reviews and Disclosures. Developers must conduct annual reviews of safety protocols and report material changes. Any misrepresentations or omissions may result in penalties.
- Enforcement and Penalties. The Attorney General may pursue civil penalties of up to $10 million per violation, or $30 million for subsequent violations.
Next Steps for Employers
Although this legislation is not yet enacted, employers and technology developers operating in New York should:
- Assess the Use of AI Systems: Identify any use of large-scale or third-party AI models that could fall within the bill’s scope.
- Review Vendor Relationships: Evaluate whether external AI tools in use are subject to the RAISE Act’s safety and transparency obligations.
- Prepare Governance Frameworks: Begin developing internal AI policies that address safety, transparency, and risk mitigation for advanced model use.
- Monitor Legislative Activity: Track the bill’s progress, as passage would introduce the first broad safety-focused AI regulation in the state.
HR Works will continue to monitor this topic and provide updates as more information is made available.
