Microsoft President Smith Wants AI Regulator
The U.S. would benefit from a new federal agency regulating artificial intelligence technology, Microsoft President Brad Smith said Thursday at a Planet Word event in Washington, D.C. Smith said the agency could oversee licensing of AI products much like regulators of automobile and aviation technology. He recommended President Joe Biden issue an executive order saying the federal government will procure AI services only from organizations applying government- and industry-sanctioned AI frameworks.
Sign up for a free preview to unlock the rest of this article
If your job depends on informed compliance, you need International Trade Today. Delivered every business day and available any time online, only International Trade Today helps you stay current on the increasingly complex international trade regulatory environment.
Conversations about AI regulation are gaining steam on Capitol Hill. Senate Majority Leader Chuck Schumer, D-N.Y., recently met with a bipartisan group of senators to discuss the topic (see 2305180050). Senate Privacy Subcommittee Chairman Richard Blumenthal, D-Conn., and ranking member Josh Hawley, R-Mo., led a recent hearing with Senate Judiciary Committee members, where OpenAI CEO Sam Altman and IBM Chief Privacy and Trust Officer Christina Montgomery urged the panel to find consensus on AI regulation.
One of the “most important questions of the 21st century” is how to regulate AI, said Smith: “This is not your typical tech person comes to Washington and says, ‘Wow, just buy our stuff and the world will be better,’” he said. “We can’t afford to look to the future that way.”
Blumenthal told reporters last week he’s participating in bipartisan discussions on drafting language for “very specific” proposals on AI regulation. The “historic feature” of the hearing with IBM and OpenAI is that industry “pleaded” with Congress to take action. “They represent a lot of the industry even though they don’t speak for others,” Blumenthal said of IBM and Open AI. The broader the regulator, the better, said Blumenthal.
It’s understandable industry wants a single AI regulator because that would make it much easier for companies to influence the regulation, said University of Pennsylvania law professor Cary Coglianese in an interview Thursday. There may be gaps that current regulators can’t account for on AI, but it might be more beneficial to have a central entity within the government to advise existing agencies on how to regulate AI, he said.
Sen. Elizabeth Warren, D-Mass., told reporters she’s “very close” to introducing legislation with Senate Judiciary Committee ranking member Lindsey Graham, R-S.C., that would establish a new tech regulator (see 2209160053). It’s important to address the “threats and benefits” of AI, and that requires having expertise from an agency that spends “24/7” concentrating on the issue, said Warren. Graham said the legislation would establish some sort of licensing mechanism for tech platforms, echoing Smith’s comments Thursday.
AI isn’t a “single product or entity,” said Coglianese. “It’s one thing to create a Nuclear Regulatory Commission because they’re going to regulate nuclear reactors. ... AI isn’t like that. AI is an application that gets used for so many different purposes, like propelling cars,” medical technology and a host of consumer products. There are “symbolic reasons” for legislators to create a new agency that will “solve all of our problems with respect to AI.” When agencies don’t solve those problems, legislators can then “beat up” on the agency, he said: That’s not to say agencies are perfect and that all legislative efforts are “theater,” but there are incentives for lawmakers to take action to make it “look like they’re solving the problem.”
Hawley criticized the FTC in conversation with reporters last week: “My experience has been, like the FTC for example, they don’t really do anything, and then they get captured by the people who they’re supposed to be regulating.” He said he has considered legislative ideas for establishing a standard of care for AI technology and a private right of action for individuals to sue when they are harmed by the technology.