Two US lawmakers have introduced bipartisan legislation aimed at establishing national standards to identify, monitor and shut down potentially dangerous Artificial Intelligence (AI) agents operating inside government and corporate computer networks. The bill, introduced on 21 September 2026, marks one of the most concrete legislative attempts yet to bring autonomous AI systems under regulatory oversight in the United States.

What the Stop Rogue AI Act Proposes

Representatives Josh Gottheimer, a Democrat from New Jersey, and Mike Lawler, a Republican from New York, are sponsoring the 'Stop Rogue AI Act', which would grant organisations greater visibility over autonomous AI systems operating on their networks. The bill would direct the National Institute of Standards and Technology (NIST) to develop national standards, guidelines and best practices for discovering, verifying, monitoring and controlling AI agents.

AI agents — software systems capable of performing tasks, making decisions and interacting with computer systems with limited human involvement — can be activated by employees, embedded in third-party software, or introduced by outside vendors, often without the host organisation's knowledge.

The Lawmakers' Case for Action

Gottheimer used a pointed analogy to frame the urgency. 'You can't drive a car in this country without a license plate, without insurance, without some way for a cop — or the person you just rear-ended — to know who's behind the wheel and who's responsible,' he said. 'AI agents right now don't have any of that. Our bill puts a driver's license on every AI agent operating in this country — so organisations can find every agent on their network, verify who built it, watch it in real time, and cut off its access the second something looks wrong.'

Lawler, meanwhile, pushed back against blanket moratoriums adopted by some states. 'A pause is not a safeguard. Some states have put a moratorium on the infrastructure that powers this technology. That does not invent a kill switch,' he said. 'If the United States taps the brakes while others floor it, we do not get safer AI. We get someone else's AI. We are not going to regulate ourselves into second place.'

Why the Stakes Are High

The lawmakers warned that organisations frequently lack a reliable method to determine how many unauthorised AI agents are active inside their systems, who created them, or what data and functions they can access. This visibility gap, they argued, becomes especially serious when AI agents operate inside hospitals, utilities, government agencies and other networks handling sensitive information and essential services. This comes amid a broader global debate over AI governance, with the European Union having already enacted its AI Act and India working on its own regulatory framework.

Three More Bipartisan Proposals

Gottheimer and Lawler also unveiled three companion measures. Their AI Workforce Training Act would provide employers with a 30% tax credit, capped at USD 2,500 per employee, for training workers in practical AI skills. The Advancing American Quantum Leadership Act would broaden the definition of quantum technology under the Export-Import Bank's China and Transformational Exports Programme. A third bill, the No Rigged Grocery Prices Act, would prohibit grocery stores and delivery platforms from using AI and personal data to charge different customers different prices for the same product.

AI and the 2026 Elections

Beyond legislation, the two lawmakers have also urged federal agencies to coordinate against AI-related threats to the 2026 elections, and have pressed major AI companies to ensure voters receive accurate and unbiased information. The dual focus — on corporate networks and democratic processes — signals that the lawmakers view unregulated AI agents as a systemic risk, not merely a technical one. How quickly Congress moves on the bill will be watched closely by the technology industry, civil society, and US allies shaping their own AI rules.