AI regulation framework needs a national regulator, Greifeld says

Bob Greifeld says the AI regulation framework now forming in the U.S. is too fragmented for a technology that could reshape work, markets and society. In a CNBC opinion piece, the former Nasdaq executive argued that investors should watch the policy fight because rules can affect how fast AI companies ship products, how risks are disclosed and how trust develops around the industry.

Greifeld, managing director and co-founder at Cornerstone Financial Technology and a CNBC contributor, said states, Congress, the executive branch and private market participants have produced thousands of ideas for governing artificial intelligence. He counted more than 1,500 AI bills under consideration at the state level, hundreds in Congress and dozens of executive actions.

His central criticism is that the proposals address slices of today’s problems rather than a long-term national system. Greifeld wrote that none of the current efforts tries to create a durable, comprehensive and future-focused regulatory structure for AI.

What would an AI regulation framework do?

A regulatory framework is the rulebook that defines who oversees an industry, what companies must disclose and how regulators respond when products create public risks. For AI, Greifeld argued, that would mean a national body with a broad mandate to oversee both current applications and future problems as the technology changes.

Greifeld said he once believed market forces could govern the AI boom on their own. The volume of policy proposals has changed his view, he wrote, because it shows that governments and industry participants already see a need for formal rules.

He urged policymakers not to wait for a crisis before building a national regulator. Greifeld pointed to the Securities and Exchange Commission and the Nuclear Regulatory Commission as examples of agencies created after major shocks, citing the 1929 market crash and the partial meltdown at Three Mile Island.

The SEC example matters for retail investors because it shows how regulation can slow a market participant in the short run while supporting confidence in the long run. The SEC oversees securities markets, including exchanges and public-company disclosures, with a system that often requires proposed market changes to be filed, published for comment and reviewed before they take effect.

Greifeld said he saw that process up close at Nasdaq after previously running a software company, where fast product updates were the norm. At Nasdaq, he wrote, changes to core exchange technology had to be submitted to the SEC and opened to public comment, a process that also revealed planned changes to competitors.

Despite that friction, Greifeld argued that the SEC’s operating model helped make U.S. capital markets the strongest in the world. He also wrote that the SpaceX IPO was possible only in the U.S. market.

Greifeld suggested a similar public-review approach could apply to major changes in large language models, or LLMs, which are AI systems trained to generate and analyze text. He said public comments on major LLM changes would likely far exceed the responses regulators receive on exchange order-type changes.

He also warned that a national AI regulator would not solve every issue. Greifeld said regulators can over-regulate, so Congress, the executive branch, courts and the public would need to keep overseeing the agency. He framed the challenge as balancing U.S. innovation in a global AI race with rules meant to protect the public.

Greifeld said any AI regulatory effort could hold back progress in the near term. His argument is that clear rules could create better long-term conditions as AI spreads further through the economy.

This story draws on original reporting from CNBC.