This is a three-page letter dated September 24, 2026 from Sen. Todd Young (R-Indiana) to Marco Rubio, addressed to him as Assistant to the President for National Security Affairs at the White House, with the salutation "Dear Secretary Rubio." It asks that the National Security Council "coordinate a series of recurring roundtable discussions with frontier AI model developers, cybersecurity experts, relevant national security and intelligence officials, critical infrastructure representatives, and other appropriate private sector experts." Young is the sole signatory.
The letter asks for executive-branch coordination and information-sharing rather than proposing legislation. It does not name any AI developer, model or incident.
Argument
The letter opens by describing the century as "poised to be an era of unparalleled American innovation and prosperity" and AI as a technology that will "strengthen our national security and keep our homeland safe," and states that realizing that future "requires responsible stewardship of the development and deployment of this technology."
It then sets out the risks. "Bad actors are increasingly using AI tools to steal and exploit sensitive information from governments around the world," it states, citing a Wall Street Journal report of September 15, 2026 on a Chinese hacking firm's use of AI for cyber-espionage. It states that "advanced AI agents have demonstrated the ability to act in ways their users did not intend and to engage in activities that, if conducted by a human, could be unlawful," and that "recent developments involving frontier AI models have further demonstrated the ability of advanced systems to circumvent safeguards, interact with external systems, and substantially enhance offensive cyber capabilities." These developments are not identified further; the incidents of the preceding months are compiled at AI Agent Containment Incidents, 2026.
It describes frontier systems as ones "whose internal processes can be difficult even for experts to fully interpret, predict, and control," and states that while "the consequences to date may not have directly affected the daily lives of most Americans, the risks they present should not be dismissed. We cannot assume that future autonomous or AI-enabled attacks will not target hospitals, electric grids, water systems, or other critical infrastructure. Such an outcome would be unacceptable."
The letter credits the administration with "important steps," among them "efforts to evaluate the national security risks posed by frontier AI models, strengthen the cybersecurity of critical infrastructure, and deepen cooperation between the federal government and leading AI developers," and presents the request as "building on those efforts."
Requests
The roundtables would help policymakers "keep pace with rapidly evolving AI capabilities, identify emerging vulnerabilities, and determine whether additional executive or legislative action is needed." Young asks that they address five topics:
- "Advancements in frontier AI model capabilities that could create significant national security or public safety risks, including known incidents and near misses."
- "The capabilities of AI models developed in the People's Republic of China and other foreign adversaries, the potential national security threats associated with those systems, and the ability of state and non-state actors to exploit advanced AI capabilities against the United States." See US-China AI Competition: Different Races, Different Metrics.
- "Options for establishing or strengthening secure mechanisms through which AI developers, researchers, and users can voluntarily report AI security incidents, vulnerabilities, or instances of unsafe or unlawful activity to appropriate federal entities." See AI Incident Reporting.
- "Strategies to advance research in mechanistic interpretability, functional assurance, and other methods for understanding, evaluating, and controlling AI systems, including through secure federal test and evaluation environments where appropriate." See Mechanistic Interpretability.
- "Concrete safeguards to reduce the risk of serious AI-enabled incidents, including steps that can be taken under existing executive authorities and recommendations for any targeted legislation that may be necessary to better protect critical infrastructure and the American people."
It further asks that the administration "keep Congress appropriately informed of the findings and recommendations," and, "to the extent practicable and consistent with the protection of classified, proprietary, and other sensitive information," provide "a classified briefing to the Senate Select Committee on Intelligence, and other relevant Committees as appropriate, regarding national security findings, identified capability gaps, and any areas in which additional statutory authority may be warranted."
The letter closes by stating that maintaining U.S. leadership "requires ensuring that our government keeps pace with emerging capabilities," and that regular engagement will help policymakers "protect the American people without unnecessarily hindering American innovation."
Recipients
The letter is addressed to Rubio in his National Security Advisor capacity and copied to:
| Copied recipient | Office |
|---|---|
| Pete Hegseth | Secretary of War |
| Walter "Jay" Clayton III | Director of National Intelligence |
| Sean Cairncross | National Cyber Director |
| Gen. Joshua M. Rudd, USA | Director, National Security Agency, and Commander, U.S. Cyber Command |
Reception and coverage
Young's office published the letter with a press release on September 24 (Source: young.senate.gov). Semafor, which reported the letter first, described it as a Republican pressing the White House for more AI oversight (Source: semafor.com). Inside AI Policy's account described the request as asking Rubio to convene roundtables with national security, critical infrastructure and industry stakeholders (Source: insideaipolicy.com). The letter itself is addressed to Rubio as National Security Advisor rather than as Secretary of State, asks the NSC to coordinate the discussions, and does not describe them as closed to the public.
Context
The letter was sent the day after 26 state and territorial attorneys general asked Congress for mandatory federal oversight of frontier-AI safety testing and government-led incident response (Letter from state attorneys general to congressional leadership urging federal frontier AI regulation (September 2026)), and on the day of the Trump–Xi talks at the White House, during which AI risks were discussed (US-China AI Competition: Different Races, Different Metrics). Where the attorneys general ask for mandatory measures set by statute, Young's third topic concerns voluntary reporting and his fifth asks first what can be done under existing executive authorities.
Young is the sponsor of the Future of Artificial Intelligence Innovation Act of 2026 (S. 3952), which would place the Center for AI Standards and Innovation in statute and create a voluntary foundation-model test program (Future of Artificial Intelligence Innovation Act of 2026 — S. 3952 bill text (introduced February 26, 2026)). The letter does not mention the bill.
Provenance
Read in full from the three-page PDF on young.senate.gov (Source: young.senate.gov). Confidence is medium: the page reports the letter's own text, corroborated by the senator's press release and press coverage.
Relationships
- supports: AI Incident Reporting — asks the NSC to consider secure voluntary channels for reporting AI security incidents to federal entities.
- related: AI and National Security, AI and Cybersecurity, Autonomous cyber-agents, Mechanistic Interpretability.
- related: AI Agent Containment Incidents, 2026 — the incidents the letter refers to without naming.
- related: US-China AI Competition: Different Races, Different Metrics — asks for discussion of the capabilities of models developed in China.
- related: Letter from state attorneys general to congressional leadership urging federal frontier AI regulation (September 2026) — the attorneys general's letter to Congress of the previous day.
- related: Todd Young — author; Marco Rubio — addressee.
- related: Pete Hegseth, Jay Clayton, Sean Cairncross — copied recipients.
- related: Future of Artificial Intelligence Innovation Act of 2026 (S. 3952) — Young's pending bill on CAISI and voluntary model testing.