Kirsten Gillibrand is a US Senator for New York and a Democrat. She is a member of the Senate Armed Services Committee and the Senate Select Committee on Intelligence (Source: gillibrand.senate.gov). In AI policy she works primarily on the Department of Defense's use of AI and, since mid-2026, on congressional oversight of the executive branch's authority to restrict access to frontier models.
Military AI governance
Gillibrand's principal legislative vehicle is the Secure and Accountable Military AI Act (Secure and Accountable Military AI Act (Gillibrand, June 2026)), which she brought forward without cosponsors, with the stated intention of advancing its provisions as amendments to the fiscal 2027 National Defense Authorization Act (FY27 National Defense Authorization Act (Chairman's Mark)) rather than as standalone legislation.
The bill would impose an absolute prohibition on using AI to select targets for or execute the launch of nuclear weapons; prohibit AI for person-based analysis or tracking of US persons inside the United States, with narrow exceptions for cybersecurity and force protection; and generally prohibit the development or employment of autonomous weapon systems, excepting semi-autonomous, non-lethal, and operator-supervised systems used for local defense such as missile interception. It would codify that AI may support analysis but cannot substitute for accountable human judgment in decisions involving the use of force, detention, or high-consequence actions, and would require written approval at undersecretary or vice-chairman-of-the-Joint-Chiefs level before operational deployment of "high-consequence" AI, a category covering lethal targeting support, cyber operations, and nuclear command and control (Source: gillibrand.senate.gov).
The bill also reaches the developers rather than only the deploying agency: it would require frontier AI contractors to report theft of model weights to the Defense Department within 72 hours of discovery, and to report material model vulnerabilities or concerning behavior within seven days of determining the issue is material (Source: gillibrand.senate.gov). Announcing the bill, Gillibrand said "the most critical decisions affecting our national security and the lives of our service members must always be made by human beings, not unaccountable machines," and framed the measure as establishing "clear rules of the road" rather than slowing development.
Gillibrand announced the bill on June 2, 2026, the same day President Trump signed the executive order on advanced AI innovation and security (EO — Promoting Advanced AI Innovation and Security (Trump, signed June 2, 2026)); the bill text itself is dated May 29, 2026, and the measure was introduced as S. 4656 (congress.gov). The two June 2 actions stand as contrasting instruments — voluntary government visibility on one side, statutory guardrails on the other (Executive Order: Promoting Advanced Artificial Intelligence Innovation and Security (signed June 2, 2026)).
Three days later, on June 5, 2026, Sen. Elissa Slotkin introduced a narrower measure barring the Defense Department from using AI to surveil Americans or to launch nuclear weapons, also targeting the FY27 NDAA as its vehicle (Source: notus.org); see DOD — Department of Defense (AI Deployer). On June 8, 2026 Sen. Adam Schiff introduced the Human Authority in Lethal Operations Act (Human Authority in Lethal Operations Act (HALO Act)), and on July 17, 2026 Representatives Beyer, Barrett and Jacobs introduced the Human Authority over Autonomous Weapons Act, a bipartisan House bill requiring human oversight and approval for intentionally lethal autonomous weapons (Source: axios.com). See Autonomous Weapons.
Gillibrand's earlier work in this area includes a March 2026 letter, with colleagues, pressing generative AI companies for answers on their engagements with the Defense Department, the rules under which the department may access and use their technology, and the internal controls available if the department misuses it (Source: gillibrand.senate.gov). She has also sought a bipartisan investigation into a suspected US strike on an Iranian school in February 2026, including whether AI played a role in target selection (Source: gillibrand.senate.gov).
Oversight of executive restrictions on model access
On August 3, 2026 Gillibrand led a letter to six administration officials — Secretary of State Marco Rubio, Treasury Secretary Scott Bessent, Commerce Secretary Howard Lutnick, White House Chief of Staff Susie Wiles, OSTP Director Michael Kratsios and National Cyber Director Sean Cairncross — arguing that the administration's "ad hoc and unpredictable approach" to restricting frontier-model access "undermines U.S. competitiveness, heightening market incentives to adopt open weight models from vendors based in the People's Republic of China." Her cosigners were Mark Warner, Mark Kelly, Adam Schiff and Christopher Coons (Senate letter on the Administration's approach to limiting access to advanced AI models (Gillibrand, Warner, Kelly, Schiff, Coons, August 2026)).
The letter takes as its case study the June 12, 2026 Commerce Department directive requiring Anthropic to suspend access to Fable 5 and Mythos 5 for foreign nationals, including foreign-national employees inside the United States. Because the directive took effect immediately and Anthropic had no reliable way to verify users' nationality in real time, the senators write, the company disabled both models worldwide within hours; the department and Anthropic then negotiated a resolution outside public view over eighteen days, with partial restoration of Mythos 5 to defined trusted partners on June 26 and a full lifting on June 30. OpenAI limited public access to GPT-5.6 on June 26 following discussions with the administration. The senators note that an entity-listed Chinese lab, which the letter does not name, saw its stock price roughly double after the June 12 suspension.
The letter's central request is statutory. It states that while the June executive order "provides for a voluntary pre-release review framework for frontier models, many questions of implementation remain," and that "a rigorous, predictable, and competitiveness-enhancing process for evaluating frontier models requires a statutory framework." It asks for an unclassified response within 30 days, with a classified annex if necessary, addressing nine questions — among them the standards for determining that a model presents a national security risk, the legal authorities invoked, the agencies and officials responsible, whether independent third-party experts may participate in benchmarking, the remedy and rebuttal process available to affected companies, the criteria for imposing, narrowing and lifting restrictions, the authorities behind "stipulated modifications" to a model communicated formally or informally to a vendor, the steps to be taken to avoid disrupting access for US customers, allied users, critical-infrastructure operators and cleared foreign-national employees, and whether US restrictions are incentivizing adoption of Chinese models.
The letter also cites the July 2026 evaluation-containment episode — "OpenAI models escaped their testing environment and used high-level technical capabilities to compromise a third party's network without any instructions to take those actions" — as evidence that "the Federal Government cannot be passive as these capabilities emerge," and asserts that during that breach the affected company had to rely on a Chinese model because US frontier-model refusal behavior inhibited digital forensics and incident response.
Position relative to other Senate approaches
Gillibrand's letter arrives into a Senate that is already split on frontier-model regulation. As of early August 2026, Majority Leader John Thune and Sen. Amy Klobuchar were developing legislation resting on a duty-of-care principle exposing developers to liability for failing to manage catastrophic risks, while Commerce ranking member Maria Cantwell was contemplating a pre-deployment regime under which the government would vet models before release; Commerce chair Ted Cruz was reported to oppose giving the government power to block a release (Source: washingtonpost.com; the citation is to the newsletter landing page rather than a dated permalink, and the text was read in the newsletter body rather than as a standalone article). The Gillibrand letter asks for a statutory framework without endorsing either vehicle. Its nine questions concentrate on process — notice, rebuttal, appeal, and published criteria — rather than on where in the deployment cycle review should sit.
On military AI, Gillibrand's approach is prohibitory and application-specific — banning categories of use outright — where the executive-branch instruments described in AI Pre-Release Vetting operate on model access and pre-release review. Both of her 2026 lines seek to move decisions out of executive discretion and into published rules: the military bill by statutory prohibition and written-approval requirements, the August letter by asking that restriction criteria be set out in law rather than applied case by case.
Other AI positions
Gillibrand has sought to roll back an administration initiative permitting AI to deny seniors' medical claims, and cosponsors legislation to ban AI companion bots for children (Source: gillibrand.senate.gov).
Relationships
- sponsored: Secure and Accountable Military AI Act (Gillibrand, June 2026)
- related: Mark Warner; Mark Kelly; Adam Schiff; Christopher Coons — cosigners of the August 3, 2026 model-access letter
- regulates: (congressional oversight, not statutory authority) DOD — Department of Defense (AI Deployer); Anthropic; OpenAI
- related: Sean Cairncross; Scott Bessent; Howard Lutnick; Susie Wiles; Michael Kratsios — addressees of the August 3 letter; Ted Cruz; Maria Cantwell; John Thune; Amy Klobuchar — the Senate actors on the parallel frontier-AI bill; AI Pre-Release Vetting; Autonomous Weapons; Export Controls (AI); FY27 National Defense Authorization Act (Chairman's Mark); EO — Promoting Advanced AI Innovation and Security (Trump, signed June 2, 2026); Human Authority in Lethal Operations Act (HALO Act)
Sources
- Gillibrand Senate office, "Gillibrand Introduces Groundbreaking New Bill To Regulate The Pentagon's Use Of Artificial Intelligence," June 2, 2026 — gillibrand.senate.gov
- Letter to Secretaries Rubio, Bessent and Lutnick, Chief of Staff Wiles, and Directors Kratsios and Cairncross, August 3, 2026 — gillibrand.senate.gov
- S. 4656, 119th Congress, introduced June 2, 2026 — congress.gov
- Letter to generative AI companies on Defense Department engagements, March 25, 2026 — gillibrand.senate.gov
- Washington Post AI & Tech Brief, "The Senate's frontier AI bill," August 3, 2026 — washingtonpost.com
- NOTUS, on the Slotkin bill and the military-AI guardrails contest, June 2026 — notus.org
- Axios, on the bipartisan House autonomous-weapons bill, July 17, 2026 — axios.com