The ROOT CAUSE Act, introduced Wednesday (Nov. 18) by Rep. Morgan Luttrell of Texas, with Rep. Ritchie Torres of New York as a cosponsor, would put the board inside the Cybersecurity and Infrastructure Security Agency, NextGov/FWC reported Wednesday.
The proposal comes after a midterm campaign in which polling showed widespread concern over AI safety. A September Quinnipiac University survey found that 86% of adults in the United States supported independent safety standards for AI companies. In the weeks before Democrats take control of Congress in January, lawmakers from both parties are pressing for a federal role in investigating critical system failures and addressing the risks posed by AI agents that can act on their own.
The proposed National Cyber and AI Investigations Board would determine what caused significant incidents and recommend ways to prevent a recurrence, the Wednesday report said. Its remit would cover government and private-sector systems, including widely used software and cloud services. The board could examine AI-assisted attacks, harmful actions by agents that exceed their permissions, and outages caused by faulty updates or configuration errors.
This month’s digital disruptions show how that jurisdiction would work. The Federal Aviation Administration imposed a nationwide ground stop Friday (Nov. 13), and a separate Microsoft outage disrupted email and other workplace tools. Officials have identified no cyberattack as the cause of either event, according to the Wednesday report.
National security systems, including the classified National Security Agency network whose outage slowed analysts’ work Tuesday (Nov. 17), would be outside the board’s jurisdiction, the report said. The NSA attributed that disruption to maintenance.
The bill would require an act of Congress to abolish, suspend or consolidate the board and bar outside officials from changing or suppressing its findings, according to the report. These protections follow the President Donald Trump administration’s removal of members of the Department of Homeland Security’s Cyber Safety Review Board while it was investigating the Salt Typhoon intrusions into telecommunications networks. The earlier board had faulted Microsoft over security failures that allowed Chinese hackers to access senior U.S. officials’ email.
Rep. Matt Van Epps of Tennessee is proposing separate safeguards for civilian agencies through the AI INSIGHT Act, the report said. CISA would have a year from enactment to establish capabilities to help agencies identify AI agents on their networks, track what they do and restrict their access to what each task requires. Agencies could shut down agents that exceed their authority and revoke access without waiting for vendor approval.
Agencies would also preserve the evidence behind AI-generated findings used in investigations or other official decisions, according to the report. Defense, intelligence and national security systems would be exempt.
The SECURE IT Act, proposed by Rep. Andy Ogles of Tennessee, would codify Homeland Security’s existing role in securing the information technology sector and assign the work to CISA, the report said. The proposal calls for closer coordination with technology providers on testing software fixes, rolling them out in stages and retaining the ability to reverse defective updates. Company participation would remain voluntary.
Sens. Maria Cantwell of Washington and Mike Rounds of South Dakota separately urged the White House to coordinate the use of AI-powered defenses and develop security standards for autonomous agents, according to a Wednesday press release. They pointed to reports that banks struggled to distinguish agents authorized to access customer accounts from unauthorized actors.
Cantwell is in line to chair the Senate Commerce Committee in January, and Rep. Bennie Thompson of Mississippi is positioned to lead House Homeland Security. The leadership change will give Democrats a greater say in how far federal oversight extends, including whether agencies and their vendors should face additional requirements for the AI agents they deploy.
For all PYMNTS AI coverage, subscribe to the daily AI newsletter.