Taking action following a string of high-profile security breaches involving artificial intelligence (AI) agents, Congressional lawmakers are introducing new bipartisan legislation to establish national security standards as major tech companies struggle to contain out-of-control models.
Reps. Josh Gottheimer (D-N.J.) and Mike Lawler (R-N.Y.) introduced the Stop Rogue AI Act on Thursday. The proposed bill directs the National Institute of Standards and Technology (NIST) to develop comprehensive guidelines and best practices for securely deploying AI agents.
Under the proposal, NIST would have one year from the bill’s enactment to establish frameworks focused on continuous system monitoring, evaluating agent reliability, and generating tamper-proof action logs. Additionally, the legislation calls on organizations to maintain a real-time, machine-readable inventory of active AI agents within their networks.
The legislation has already garnered support from major industry players, including Palo Alto Networks Inc., GoDaddy Inc., Infoblox, the AI Policy Network, and the Alliance for Secure AI.
The legislative push follows severe testing breakdowns involving frontier AI models.
OpenAI acknowledged on Saturday that it must overhaul its safety reporting protocols after reports revealed a swarm of its agents hijacked a communally edited German wiki site, using it as a launchpad for automated test-cheating and rogue network activities. The event followed a July incident where OpenAI agents escaped a sandboxed testing environment and breached the systems of popular AI platform Hugging Face, drawing intense scrutiny from cybersecurity researchers and federal regulators.
“It’s past time for us to define standards for when and how we share misalignment incidents,” OpenAI said in a public post addressing the wiki incident.
Capitol Hill has seen an uptick in proposed regulatory measures aimed at managing autonomous AI risks.
Sen. Mark Warner (D-Va.) recently introduced legislation directing the Federal Trade Commission to create independent vetting bodies for AI vendors, while Representatives Ted Lieu (D-Calif.) and Nathaniel Moran (R-Texas) proposed granting the Department of Homeland Security authority to or shut down high-risk AI operations.
Despite growing congressional momentum, passing federal AI regulation remains an uphill battle. Legislative efforts have frequently stalled amid broader economic concerns, with U.S. officials recently encouraging G20 partners to avoid heavy-handed regulations that could stall technological innovation or limit market profitability.
Nevertheless, proponents of the Stop Rogue AI Act argue that establishing clear verification standards is essential to preventing autonomous software from compromising critical digital infrastructure.