New York Laws “RAISE” the Bar in Addressing AI Safety: The RAISE Act and AI Companion Models
The state of New York was a leader in artificial intelligence (AI) regulation in 2025. Among its significant actions, the state legislature enacted an omnibus budget law implementing safeguards for AI companions, effective November 5, 2025. Furthermore, the Governor signed an amended version of the Responsible Artificial Intelligence Safety and Education (RAISE) Act on December 19, 2025. New York now joins states like Colorado, California, and Utah in a growing framework of comprehensive and targeted AI legislation.
As states push forward with their regulatory efforts, federal activities have raised questions regarding pre-emption and the future of state-level AI legislation. For instance, an executive order (EO) issued in December directed the creation of an AI Litigation Task Force within the DOJ aimed at challenging state laws. This EO has sparked debates about federal authority over state regulations, introducing a period of legal uncertainty that could influence states’ regulatory timelines.
Artificial Intelligence Companion Models
The AI Companion Model law mandates operators and providers of AI companions to implement safety measures addressing users’ expressions of suicidal ideation or self-harm. It also requires regular notifications to users that they are not interacting with a human being.
Scope
This law applies to all operators of AI companions with users located in New York. AI companions are defined as systems using AI, generative AI, and/or emotional recognition algorithms designed to foster human-like relationships with users. Key features include retaining user interaction history, asking unsolicited emotional questions, and maintaining ongoing dialogues on personal matters.
Key Requirements
Operators must provide clear notifications to users, either verbally or in writing, about the non-human nature of the interaction. This notification should occur at least every three hours during ongoing interactions. Moreover, operators are required to implement protocols to identify and address suicidal ideation or self-harm, referring users to appropriate crisis services when necessary. Violations of these provisions may result in civil penalties of up to $15,000 per day, enforced by the state attorney general.
Next Steps
Operators covered by this law need to establish how to effectively notify users and ensure the clarity and frequency of these notifications. They must also develop methods to detect and address expressions of suicidal ideation or self-harm.
New York’s approach to AI companions is novel, following California’s similar law effective January 1, 2026. California’s legislation shares many requirements with New York’s, particularly regarding notifications and protections for minors.
Responsible Artificial Intelligence Safety and Education (RAISE) Act
Alongside the AI Companion Model law, New York tackled broader AI safety issues through the RAISE Act, effective January 1, 2027. This act mandates developers of frontier models to implement transparency and disclosure requirements, including making safety protocols available to relevant authorities.
Key Requirements
The RAISE Act applies to frontier models developed, deployed, or operated in New York. These models are defined as large-scale AI systems requiring extensive computational resources and significant financial investment. Key requirements for developers include:
- Conducting annual safety reviews and independent third-party audits.
- Publishing safety protocol information while allowing some redactions.
- Reporting safety incidents within 72 hours.
- Determining whether models could cause “critical harm.”
- Creating detailed safety and security protocols to prevent critical harms.
The current version of the RAISE Act prohibits the deployment of models posing an unreasonable risk of critical harm. Violations may result in civil penalties ranging from $1,000,000 for the first offense to $3,000,000 for subsequent violations.
Conclusion
New York has enacted several AI-related bills in 2025, addressing various aspects such as personalized algorithmic pricing and digital replicas. With over a dozen states introducing AI-specific laws, the landscape is rapidly evolving. Despite the uncertainty created by recent federal actions, further state-level proposals are anticipated in 2026.
Stay tuned for forthcoming recommendations for compliance with new AI laws set to take effect in 2026.
