
Two senior US senators have warned that artificial-intelligence tools are increasing the speed and scale of cyber threats, calling for a coordinated national effort to protect critical infrastructure, financial institutions and widely used software. Senator Maria Cantwell, a Democrat from Washington, and Senator Mike Rounds, a Republican from South Dakota, have urged federal authorities to strengthen the country’s ability to detect and respond to attacks that use AI to identify software weaknesses and automate parts of the exploitation process. The November 18 announcement describes their call for urgent White House action.
In a letter dated November 17, the senators told the Super Intelligence Task Force that AI was rapidly increasing attackers’ ability to discover and exploit software weaknesses, while defensive systems were not keeping pace. They cited concerns involving vulnerabilities in widely used software, unauthorised AI agents accessing financial accounts and cyber incidents affecting organisations in Japan. The letter also raised questions about the security implications of foreign-developed AI models, while making clear that one cited test did not establish deliberate malicious behaviour. The senators’ official letter sets out the examples behind their warning.
The lawmakers’ concerns come as security researchers report growing use of autonomous AI frameworks in cyber operations. Google’s Threat Intelligence Group has described attackers using large language models to reverse-engineer software patches and automate vulnerability exploitation. Such capabilities can shorten the time available for organisations to install fixes, particularly where systems are complex or dependent on legacy software. Google’s November threat assessment describes how AI is being integrated into reconnaissance, exploitation and other stages of cyber intrusions.
The senators are pressing for a response built around stronger standards, testing, auditing and defensive AI. Their proposal calls for government agencies, universities, technology companies and critical-infrastructure operators to work together to develop and validate tools that can identify threats, strengthen vulnerable systems and accelerate patching. They have also argued that independent testing is needed to evaluate the security behaviour of advanced AI systems. The objective is not simply to restrict AI development, but to ensure that the systems being deployed are tested and that the same technology can be used to improve cyber defence. A previous Senate committee statement outlines the lawmakers’ broader recommendations.
The warning raises a difficult policy challenge for Washington: AI is becoming a tool for both attackers and defenders, and the pace of development is faster than many traditional regulatory and procurement processes. Critical infrastructure operators often rely on interconnected systems that cannot be updated instantly without careful testing. A national response will therefore require practical coordination, clear security standards, rapid information-sharing and support for organisations that lack large cybersecurity teams. The senators’ appeal adds pressure on federal agencies to turn growing concern about AI-enabled attacks into concrete protective measures. The full letter is available through the Senate Committee on Commerce, Science and Transportation.