New York is looking to make artificial intelligence safety violations considerably more expensive.
A proposed expansion of the state’s AI safety legislation could expose major developers to penalties reaching $3 million per day for repeated violations. The measure, known as RAISE Act 2.0, would also introduce independent safety examinations. It would create stricter reporting requirements and greater regulatory control over potentially dangerous AI systems.
State Senator Andrew Gounardes introduced the proposal in October 2026. He built on New York’s existing Responsible AI Safety and Education (RAISE) Act.
The timing is significant. AI developers are moving quickly toward systems capable of writing software, conducting research and performing tasks with limited human supervision. New York lawmakers want stronger safeguards before those capabilities become even more widespread.
New York AI Safety Fines Could Reach $90 Million in One Month
Under New York’s existing RAISE Act, developers face penalties of up to $1 million for an initial violation and $3 million for subsequent violations. RAISE Act 2.0 would change how those penalties accumulate. As a result, regulators could pursue fines of up to $1 million per day for a first violation and $3 million per day for repeated violations. A continuing violation lasting 30 days could therefore result in a maximum penalty of $90 million.
The proposed penalties would not apply automatically. Enforcement would require legal action, and the amount imposed would depend on the circumstances of each case. Still, the change would create a much stronger financial incentive for developers to resolve safety failures quickly. What previously represented a fixed regulatory penalty could become a growing liability for every day a company remains out of compliance.
Independent AI Safety Audits Would Become Mandatory
New York also wants to reduce reliance on AI companies evaluating their own safety practices. Under the proposed legislation, covered developers would appoint independent, state-licensed examiners to review compliance with safety requirements. These examiners would conduct annual audits and assess catastrophic risks every 90 days. Their findings would be submitted to regulators and disclosed publicly.
The proposal would place additional responsibility on company executives. Senior officers would need to certify regulatory submissions under penalty of perjury, while developers would provide quarterly information about their use of AI systems in research and development. For companies building increasingly powerful models, independent scrutiny could become a routine part of operations. This would replace the current practice of occasional responses to public concern.
AI Developers Spending Over $1 Billion Could Face Regulation
The proposed legislation would also expand which companies fall under New York’s AI safety requirements. The original RAISE Act focuses on developers exceeding a $500 million annual revenue threshold. RAISE Act 2.0 would extend coverage to developers spending more than $1 billion on frontier AI development over a three-year period.
That distinction could matter for emerging AI laboratories backed by substantial investment but generating relatively little revenue. Some developers spend enormous amounts on computing infrastructure, model training and research long before their products become profitable. By considering development expenditure, New York would be able to regulate more companies based on the scale of their AI activities. This approach goes beyond simply considering their commercial income alone.
New York Wants Greater Control Over High-Risk AI Systems
Financial penalties are only one part of New York’s proposed approach. The legislation would give state regulators additional authority to address advanced AI systems that present serious safety concerns. This could include restrictions on computing resources used for model training or automated AI research. Additionally, the attorney general could seek court orders limiting certain dangerous activities.
The proposal also addresses incidents involving AI systems operating beyond their intended boundaries. Developers could face expanded reporting obligations when models bypass safety controls, conduct unauthorized cyber activities or narrowly avoid serious failures. These concerns have become more pressing as AI agents gain access to coding environments, external tools and increasingly complex workflows.
Businesses Remain Cautious About Autonomous AI Agents
Businesses are adopting AI agents, but many remain reluctant to give them unrestricted operational authority. Research cited by PYMNTS Intelligence suggests that enterprises increasingly use agentic AI while maintaining limits on what these systems can access and execute without human approval.
That caution reflects a practical concern. An AI assistant answering customer questions presents different risks from an autonomous system modifying software, handling sensitive information or interacting with external networks. New York’s proposed legislation focuses primarily on developers of advanced AI models. However, stricter safety requirements could eventually influence the products and services those developers offer enterprise customers.
New York’s AI Regulation Push Could Influence Other States
New York’s proposal comes as American states develop their own approaches to AI oversight. California has introduced safety requirements for frontier AI developers, while other jurisdictions are exploring transparency obligations, independent assessments and incident reporting rules.
RAISE Act 2.0 would add another layer to that evolving regulatory landscape. Supporters see stronger enforcement as necessary to ensure developers take safety commitments seriously. For AI companies, the proposed requirements could mean additional compliance costs and more external scrutiny. There could also be potentially greater restrictions on how advanced systems are developed and deployed. The debate is increasingly moving beyond whether AI needs oversight to how much authority regulators should have over the companies building it.
What RAISE Act 2.0 Means for the Future of AI Safety
RAISE Act 2.0 has not yet become law, and its provisions could change as the legislative process continues. New York’s existing RAISE Act is scheduled to take effect in January 2027. This makes the proposed expansion particularly relevant for developers preparing to comply with the state’s initial requirements.
The $3 million daily penalty may attract the headlines, but the independent audits and expanded enforcement powers could have longer-lasting implications. New York is proposing a system in which AI safety claims face greater external scrutiny, and unresolved violations carry potentially substantial financial consequences.
For developers racing to build more capable AI systems, the message is becoming harder to ignore: safety requirements may soon carry costs that extend far beyond reputational damage.
Sources
1. PYMNTS — New York Proposes $3 Million Daily Fines for AI Safety Violations (November 11, 2026).
2. New York State Senate — Senate Bill S10731 — RAISE Act Amendments (Official legislative text and status).
3. New York State Senate — Sen. Gounardes Proposes Major New AI Safety and Transparency Regulations (October 29, 2026).
4. TechCrunch — OpenAI, Anthropic Face New Oversight as New York’s AI Safety Law Takes Effect (November 10, 2026).
