Estimated read time: 3-4 minutes
WASHINGTON — Utah Sen. Mitt Romney and several Senate colleagues on Tuesday announced an outline to protect against the potential risks posed by artificial intelligence — including biological, chemical, cyber and nuclear threats — by enhancing federal oversight of future cutting-edge developments in the technology.
The bipartisan group of senators — including Romney, a Republican, along with Sens. Jack Reed, D-Rhode Island, Jerry Moran, R-Kansas, and Angus King, I-Maine — outlined their framework in a letter to leaders of the Senate's artificial intelligence working group. The plan addresses so-called frontier models yet to be developed, and would only apply to models designed with a broad range of capabilities or "are intended to be used for bioengineering, chemical engineering, cybersecurity or nuclear development," according to a fact sheet.
"AI has the potential to dramatically improve and transform our way of life, but it also comes with enormous risks to national security and our humanity at large," Romney stated. "My colleagues and I have spent the last several months developing a framework which would create safeguards and provide oversight of frontier AI models aimed at preventing foreign adversaries and bad actors from misusing advanced AI to cause widespread harm.
"It is my hope that our proposal will serve as a starting point for discussion on what actions Congress should take on AI — without hampering innovation."
Although artificial intelligence only emerged in the public conscience with the release of ChatGPT less than two years ago, policymakers across the globe are already grappling with what guardrails, if any, should be placed on the rapidly improving technology. The European Union approved the first major set of regulations last month, placing artificial intelligence uses deemed to pose a higher threat under increased scrutiny across the 27-nation bloc.
U.S. leaders have yet to adopt comprehensive AI regulations, but Congress has held several panels and sessions with tech leaders, such as OpenAI CEO Sam Altman, to address the technology. Romney said he is "in the camp of being more terrified about AI than I am in the camp of those thinking this is going to make everything better for the world," during one such Senate hearing last September.
President Joe Biden in October signed an executive order focused on AI safety and privacy for Americans.
AI regulation is also a concern at the state level, and Utah lawmakers earlier this year created the Artificial Intelligence Learning Laboratory to help keep the state government abreast of the latest opportunities and risks created by advancing technology.
The senators' letter for AI regulation contains few specifics, instead outlining a broad regulatory framework to follow. Similar to the European regulations, the framework proposes a tiered licensing structure where high-risk artificial intelligence models "could be licensed for deployment with vetted customers or limited public use," rather than being deployed generally.
The proposal recommends several options for oversight of AI, including the Department of Commerce's National Institute for Standards and Technology and Bureau of Industry and Security, or the U.S. National Laboratories that is housed in the Department of Energy. Congress could also choose to create a new federal agency or new interagency coordinating body to work between existing agencies.
Regardless of who oversees the regulations, the senators recommend that any oversight entity include "subject matter experts who could be detailed from relevant federal entities," and "skilled AI scientists and engineers."
Romney and his colleagues, in the letter, acknowledge their framework is only meant to "serve as the basis for international coordination to mitigate extreme risks posed by AI," and they encourage further dialogue from congressional leaders and other stakeholders "in the weeks to come, which will be necessary for us to create a framework that can become law."
