The U.S. has 1,200 AI bills and no good test for any of them

In a recent interview, IBM CEO Arvind Krishna highlighted the urgent need for balanced AI regulation in the U.S., amidst a surge of over 1,200 AI-related bills introduced across state legislatures. He warned that a bloated regulatory framework could stifle innovation and hinder the country's competitive edge in the global AI landscape. With states like California, New York, and Texas pushing varied legislation aimed at transparency and safety, the lack of a cohesive federal policy complicates the situation, risking a fragmented approach that could increase compliance burdens without effectively addressing existing legal protections.

Krishna's concerns underscore the potential pitfalls of oversimplifying the debate to a binary choice between heavy regulations and unchecked deployment. A proposed three-stage test to improve legislative clarity advocates for evaluating whether existing laws already cover the intended protections, weighing the cost-benefit analysis of new rules, and ensuring targeted legislative efforts address specific harms without creating unnecessary barriers. As countries like the EU and China advance their AI regulatory frameworks, the U.S. risks falling behind if it cannot establish a clear and coherent policy that balances innovation, safety, and national security.