Tensions between individual states and the Federal government over how to regulate artificial intelligence (AI) are starting to increase in the wake of legislation signed into law in New York and concerns raised by Ron DeSantis, governor of Florida.
Democratic Governor Kathy Hochul of New York has signed legislation that requires developers of AI frameworks for frontier models to create and publish information about their safety protocols, and report incidents to the state within 72 hours of determining that an incident occurred. An extension to an existing Responsible AI Safety and Education (RAISE) Act, the state of New York has now created an oversight office within the Department of Financial Services that will assess developers of large frontier to provide greater transparency via an annual report.
The Republican governor of Florida, meanwhile, said that AI parental rights and protections for kids and parents will be a legislative priority in the New Year.
U.S. President Donald Trump, meanwhile, has previously issued an “Ensuring a National Policy Framework for Artificial Intelligence” executive order that seeks to limit what is described as excessive and cumbersome AI regulation at the state level. The concern is that those regulations might inhibit or slow the pace of AI research and development during what has become a global race to lead the advancement of AI.
The degree to which any state regulation pertaining to safety might impact the pace of AI development is, however, debatable, says Futurum Group president and COO Dan O’Brien. “This is all about creating transparency, while allowing innovation to flourish,” he says. “I don’t see the safety reporting requirements as burdensome.”
In fact, the safety regulations being applied by New York are somewhat similar to what companies like Anthropic already publish with its Transparency Hub, noted O’Brien. The after-the-fact reporting is more meaningful, and represents an approach where new regulation could have some teeth in terms of driving real accountability, he added.
However, given the recent executive order from President Trump aimed at preempting conflicting state laws it is not clear how enforceable any state regulation may be, said O’Brien.
Most states have at least one regulation that pertains to how AI is allowed to be used. Less clear is to what degree the U.S. Congress might craft national AI legislation that could overcome a probable presidential veto. In the meantime, however, countries that initially adopted more restrictive AI policies are now moving to loosen some of those regulations in the interest of accelerating the pace of innovation.
Most IT professionals should assume that despite what some proponents of AI might be hoping, there will be more AI regulations being enacted in the months ahead. The issue, as always, is striking a balance between safety and innovation. Otherwise, the pace at which AI is embraced is only going to be slowed due to a simple lack of trust in an emerging technology that, while amazing, raises a host of societal issues that have yet to be addressed.
