The votes are in for the 2025 FedScoop.
By
Federal agencies would be required to develop artificial intelligence standards and use the National Institute of Standards and Technology’s AI guidelines under a bipartisan bill introduced Thursday.
Led by Rep. Ted Lieu, D-Calif., the bill would require agencies to use the Artificial Intelligence Risk Management Framework, developed by the NIST in 2023, and work with the agency in developing other consistent standards and guidelines.
Reps. Zach Nunn, an Iowa Republican, and Don Beyer, a Virginia Democrat, co-sponsored the bill, with Beyer calling it “a natural starting point” to ensure agencies have the tools they need to navigate AI’s complexities.
“This bill lays the foundation for harnessing the power of AI for the benefit of the American people, while upholding the highest standards of accountability and transparency,” Beyer said in a statement.
The bill would also direct NIST to recommend training and use the standards when acquiring any AI systems or services. The guidelines would not apply to national security systems.
Originally designed as voluntary, the framework provides a structure to manage AI risk and promote responsible use through four core principles: govern, map, measure and manage.
“Understanding and managing the risks of AI systems will help to enhance trustworthiness, and in turn, cultivate public trust,” it says.
The framework was created to fulfill a requirement under the National AI Initiative Act, part of the enacted fiscal 2021 National Defense Authorization Act.
Lieu, who previously co-led the House’s Bipartisan Artificial Intelligence Task Force, also sponsored legislation last month that combines bipartisan bills resulting from task force recommendations and a report. The task force’s report also said the NIST framework “only sets the theoretical baseline … and so, is not a standards document.”
“Congress should support NIST in developing additional voluntary guidance and resources for federal agencies adopting AI systems and promulgating AI-related standards,” the 2024 report stated. “For example, NIST should consider developing a risk profile of the AI risk management framework for specific federal systems.”
While NIST said it does not comment on pending legislation, the institute announced last month that it is currently developing a new profile for the framework on trustworthy AI in critical infrastructure.
The bill has been referred to the House Science, Space and Technology Committee for further action.
Federal agencies would use NIST’s AI guidelines under bipartisan House bill – FedScoop
Leave a Comment
