US to Set Up New AI Safety InstituteUK-US Will Partner on Monitoring Advanced AI Before Its Release
The U.S. government will set up a dedicated artificial intelligence safety institute to lead cooperation among the public and private sectors to develop secure AI systems, a top Biden administration official said.
Speaking at the U.K. government's AI Safety Summit on Wednesday, U.S. Secretary of Commerce Gina Raimondo said the AI Safety Institute will be housed at the Department of Commerce.
The AI Safety Summit, currently underway in Bletchley Park in the U.K., intends to focus on preventing the misuse of emerging AI capabilities that are deemed dangerous enough to pose "severe risks to public safety." The guest list includes U.S. Vice President Kamala Harris, Google DeepMind CEO Demis Hassabis, and Tesla CEO Elon Musk.
Raimondo said Commerce headquarters will work with the National Institute of Standards and Technology to develop "best in class standards for AI safety, security and testing."
"This institute will also be providing testing environments to evaluate known risks and emerging risks of AI at the frontier," she added.
The announcement from the U.S. government comes just days after the Biden administration unveiled its AI executive order that directs developers of foundational models to notify the government and share safety test results. Raimondo said the institute builds on the executive order, as well as voluntary commitments signed by 15 leading AI companies in the U.S.
The U.S. government's latest initiative follows a similar measure unveiled by British Prime Minister Rishi Sunak, who last week announced the United Kingdom will set up a dedicated AI Safety Institute to lead artificial intelligence innovation in the country.
The primary goal of the new institute will be to "ensure AI systems are safe before they are released," Sunak said at the time.
Addressing the audience on Wednesday, Raimondo said the U.K. and U.S. governments will establish a formal relationship between the two new institutes.
"Finally, of course, the work does not begin and end with just the U.S. and the U.K. We want to expand information sharing, research, collaboration and, ultimately, policy alignment across the globe," she said.
In October, a European Union lawmaker said the trading bloc will stand up a dedicated "AI office" to coordinate the compliance requirements under the proposed EU AI Act, which is likely to be finalized toward the end of this year (see: EU Will Stand Up Office to Enforce AI Act, Says EU Lawmaker).