House bill would task NSA with developing AI security playbook to counter China
Rep. Darin LaHood, R-Ill., leaves a meeting of the House Republican Conference in the U.S. Capitol on Tuesday, September 24, 2024. LaHood introduced legislation with colleagues from both sides of the aisle June 12 that would task the NSA with developing guidance on how to protect AI systems from threat actors. Tom Williams/CQ-Roll Call, Inc via Getty Images
Lawmakers want the NSA to develop guidance to help companies better protect the development of their advanced AI technologies from threat actors.
Lawmakers are tasking the NSA with drafting a national AI security strategy to keep sensitive technologies out of the hands of foreign adversaries.
Bipartisan House members introduced legislation on Thursday that would require the National Security Agency to create an artificial intelligence “security playbook” to protect sensitive U.S. technologies from foreign adversaries like China.
The bill was introduced by Rep. Darin LaHood, R-Ill., and co-sponsored by Reps. John Moolenaar, R-Mich., Raja Krishnamoorthi, D-Ill., and Josh Gottheimer, D-N.J. Moolenaar is the chairman of the Select Committee on China and Krishnamoorthi is the panel’s ranking member.
In a press release, the lawmakers said there was evidence that Chinese-based startup DeepSeek’s AI chatbot “used illegal distillation techniques to steal insights from U.S. AI models to accelerate their own technology development,” and that the legislation was needed “to address vulnerabilities, threat detection, cyber and physical security strategies, and contingency plans for highly sensitive AI systems.”
LaHood and Gottheimer previously introduced a bill in February to prohibit federal employees from downloading DeepSeek’s app. Their proposal came after security researchers voiced concerns about DeepSeek’s ties with the Chinese Communist Party, as well as the app’s ability to send user data to Chinese-owned telecommunications provider China Mobile.
The legislation would require the NSA’s AI security playbook to include information on potential vulnerabilities “in advanced AI data centers and among advanced AI developers capable of producing covered AI technologies,” with a particular emphasis on the cybersecurity risks “and other security challenges that are unique to protecting covered AI technologies and critical components of such technologies.”
To counter adversaries like China, the proposal would also direct the NSA to identify AI components or information that, if accessed by nefarious actors, “would meaningfully contribute to progress made by the actor with respect to developing covered AI technologies.”
With a focus on safeguarding the development of critical AI systems, the guidance would also have to include strategies for firms to detect and respond to cyber threats, as well as produce an analysis of how the U.S. government would need to be involved in AI-related development and oversight.
“As AI becomes an integrated part of society, Congress must ensure that the United States fortifies our technology defenses and remains the undeniable leader,” LaHood said in a statement, adding that the legislation would “draw a clear line that we will not allow our foreign adversaries to steal, exploit, or weaponize American innovation.”
The bill’s introduction comes as U.S. officials and lawmakers have increasingly expressed concern about China, in particular, gaining access to key American-made components to fast-track its development of competing AI technologies.
During a panel discussion at the AWS Summit on Tuesday, White House Artificial Intelligence and Cryptocurrency Czar David Sacks said the U.S. was focusing on policies and regulations that would help the country outpace China’s AI effort.
“It is a race,” Sacks said during the event. “It's a race to build out infrastructure. It's a race to build out applications. It's a race to proliferate those applications. It’s a race to get users. And we want to win that race. We want to get our companies to win that race.”