Coalition Urges New York Lawmakers to Avoid Heavy-Handed AI Mandates
Opposing Assembly Bill A. 6453 / S. 6953 (RAISE Act)
Dear Speaker Heastie, Assembly Minority Leader Barclay, Senate Majority Leader Stewart-Cousins, and Senate Minority Leader Ortt:
We, the undersigned organizations, write to express profound concerns regarding Assembly Bill A. 6453 / S. 6953, the “Responsible AI safety and education act” (RAISE Act). While we share the goal of ensuring the safe and equitable distribution of artificial intelligence (AI) innovations, A. 6453 / S. 6953 would create substantial obstacles to innovation, impose unworkable burdens on technology developers, and negatively impact the broader technology ecosystem, both within New York and nationally.
The RAISE Act will harm New York’s vibrant and growing tech sector
A primary concern arises from the bill’s imposition of a prophylactic, pre-deployment regulatory regime on developers of generative AI models. Mandating comprehensive reviews and testing protocols, namely the annual independent audits mandated in Article 44-B, § 1421(4) of the bill, as well as potentially requiring armative approval before a model can be introduced or even significantly updated, introduces considerable friction and delay into the innovation cycle. Such requirements function as significant barriers to entry, disproportionately aecting upstart model developers that lack the extensive resources necessary to navigate complex, front-loaded compliance processes.
This environment risks stifling the dynamism of the AI industry in its infancy and discouraging new market entrants with potentially groundbreaking approaches. New York is home to more than 2,000 AI startups, including 35 “unicorns” that have collectively raised over $17 billion as of last year [1]. They exemplify the kind of innovation at risk: agile, creative firms pushing the boundaries of what’s possible in human-computer interaction. Among other things, these startups are building models that help scientists accelerate drug discovery, enable educators to personalize learning at scale, and empower cities to plan smarter infrastructure using real-time data. The potential consequence is a slowdown of the entire AI stack, limiting the emergence of novel applications and beneficial tools that rely on these underlying technologies. The iterative nature of AI development, where progress often occurs through rapid testing and refinement, is fundamentally misaligned with A. 6453 / S. 6953’s rigid pre-deployment framework. Moreover, consumers may ultimately be less protected if model developers are spending excessive time and resources on pre-deployment compliance checklists and fixed documentation criteria that may be obviated by technological progress.
Additionally, the RAISE Act establishes an exceptionally broad and potentially unmanageable liability standard for the developers of AI models. The legislation holds model developers responsible for harms arising from the misuse of AI tools by third parties, even when those tools are applications built atop the developer’s original model. Thus, a foundational model developer must anticipate and account for the myriad ways third parties might misuse the technology. Given the general-purpose nature of AI models and the sheer unpredictability of human ingenuity, this is an untenable expectation.
Further, by tying fines to a company’s total compute spend and personnel headcount, A. 6453 / S. 6953 penalizes the very resources that drive AI research and development. A developer cannot reasonably foresee, let alone control, every possible misuse made by others. Holding developers liable for unpredictable or theoretical harms is like holding a car manufacturer responsible for every accident on the road. It creates legal and financial uncertainty that would significantly deter investment in AI development.
Finally, the RAISE Act is premature. Any sweeping new regulatory proposal must enunciate a clear goal, an underlying policy rationale, and an armative case that the benefits will outweigh the costs of compliance and foregone innovation. A. 6453 / S. 6953 comes up short on all three. In sum, proponents have not demonstrated that this overarching new regulatory structure will meaningfully advance AI safety.
We also note that A. 6453 / S. 6953 exempts EmpireAI, an ambitious project to unlock innovation through the state’s world-class universities. While we support EmpireAI, exempting it begs the question: if models are inherently dangerous, as this bill implies, then why exempt any of them? In essence, leaving it out implicitly acknowledges that this expensive regulatory eort is unnecessary.
In contrast to its speculative benefits, the harms of the RAISE Act are all too real
The RAISE Act represents a significant expansion of regulatory oversight with uncertain benefits but clear costs to the innovation economy, potentially diverting critical resources from research and development toward compliance activities. Perhaps most concerning is the bill’s apparent extraterritorial reach, seeking to regulate model developers nationwide, irrespective of their physical location or direct business ties to New York. This provision eectively positions New York as a de facto national regulator for AI development, creating potential conflicts with other state or federal frameworks and establishing a precedent that could significantly chill innovation nationwide. Companies may hesitate to deploy advanced AI tools here, ultimately disadvantaging consumers and businesses within the state by limiting access to beneficial technologies.
A more eective path forward involves targeted regulatory eorts focused on specific misuse of AI, rather than broad, pre-emptive regulation of foundational models. Fostering industry best practices, supporting research into AI safety and alignment, and enforcing existing laws against harmful misuse of technology oer a more balanced approach that will protect consumers without cutting them o from valuable AI innovations.
The RAISE Act risks undermining New York’s leadership position in technology and innovation. We urge you to reconsider this bill and its consequences.
For that reason, we oppose A. 6453 / S. 6953.
Thank you for your time and consideration of these critical issues.
Sincerely,
Chamber of Progress
Software & Information Industry Association
Business Software Alliance
Computer & Communications Industry Association
Engine
Neil Chilson, Head of AI Policy, Abundance Institute*
NetChoice
R Street Institute
TechFreedom
* – Organization affiliation provided for identification purposes only and does not imply endorsement by the organization itself.
[1] Tech:NYC. NYC Tech Snapshot 2025. https://www.technyc.org/nyc-tech-snapshot-2025