[Congressional Bills 119th Congress]
[From the U.S. Government Publishing Office]
[S. 4113 Introduced in Senate (IS)]
<DOC>
119th CONGRESS
2d Session
S. 4113
To provide for limitations on the use of artificial intelligence by
Department of Defense.
_______________________________________________________________________
IN THE SENATE OF THE UNITED STATES
March 17, 2026
Ms. Slotkin introduced the following bill; which was read twice and
referred to the Committee on Armed Services
_______________________________________________________________________
A BILL
To provide for limitations on the use of artificial intelligence by
Department of Defense.
Be it enacted by the Senate and House of Representatives of the
United States of America in Congress assembled,
SECTION 1. SHORT TITLE.
This Act may be cited as the ``AI Guardrails Act of 2026''.
SEC. 2. LIMITATIONS ON USE OF ARTIFICIAL INTELLIGENCE BY DEPARTMENT OF
DEFENSE.
(a) Sense of Congress.--It is the sense of Congress that,
consistent with America's AI Action Plan dated July 2025, the United
States must aggressively adopt artificial intelligence (AI) within its
Armed Forces if it is to maintain its global military preeminence while
also ensuring that the Department of Defense's use of AI is secure and
reliable.
(b) Limitations.--The Department of Defense is prohibited from
using artificial intelligence as follows:
(1) For the execution of launching or detonating a nuclear
weapon.
(2) For the monitoring, tracking, profiling, or targeting
of individuals or groups in the United States, without an
individualized, articulable legal basis, regardless of the
origin of the data used. In no event may the Department of
Defense use AI solely for the purpose of monitoring, tracking,
profiling, or targeting activities protected by the First
Amendment or the lawful exercise of other rights secured by the
Constitution or laws of the United States.
(3) In the employment of lethal force by autonomous weapon
systems without appropriate levels of human judgment and
supervision. All other uses of artificial intelligence in
autonomous weapon systems shall be consistent with Department
of Defense Directive 3000.09, ``Autonomy in Weapon Systems,''
dated January 25, 2023.
(c) Waiver for Certain Autonomous Weapon Systems.--
(1) In general.--The Secretary of Defense, without
delegation, may waive the prohibitions under subsection (b)(3)
with respect to a system for up to one year, or renew such a
waiver for up to one year, if the Secretary certifies, in
writing, to the congressional defense committees that
extraordinary circumstances affecting the national security of
the United States require the waiver and that the probability
of the system producing a result inconsistent with commander
intent does not exceed the documented error rate of trained
human operators performing equivalent functions under
equivalent conditions.
(2) Notifications.--For each waiver issued under paragraph
(1) with respect to a system, the Secretary of Defense shall
notify Congress, including submission of the certification
required under such paragraph, not later than 5 days after--
(A) the issuance of a waiver for formal development
of such system;
(B) the issuance of a waiver for fielding of such
system; and
(C) any modification to such system that results in
the system algorithms, intended missions sets, intended
operational environments, intended target sets, or
expected adversarial countermeasures to substantially
differ from those granted for previously waived uses.
(3) Elements.--Each notification submitted under paragraph
(1) shall include the following elements:
(A) The rationale for the waiver.
(B) A description of the autonomous weapon system
or technology covered by the waiver.
(C) A description of the operational parameters and
safeguards for the system, including:
(i) Assessments of system performance,
capability, reliability, effectiveness, and
suitability under realistic conditions.
(ii) A description of the associated
training, doctrine, and tactics, techniques,
and procedures.
(iii) The timeframe and geographic area of
intended use.
(iv) A description of measures taken to
minimize the probability and consequences of
unintended engagements or failures.
(v) Clear procedures for trained operators
to activate and deactivate system functions.
(vi) Post-deployment continuous monitoring
mechanisms.
(vii) Results from the realistic system
developmental and operational testing and
evaluation that demonstrate the probability of
the system misidentifying a target, taking
unintended action, or producing a result
inconsistent with commander intent does not
exceed the documented error rate of trained
human operators performing equivalent functions
under equivalent conditions.
(D) The anticipated duration of the waiver.
(4) Form.--A notification under paragraph (1) shall be
submitted in unclassified form, but may include a classified
annex, as the Secretary determines necessary.
(d) Artificial Intelligence Defined.--In this section, the term
``artificial intelligence'' has the meaning given such term in section
5002 of the National Artificial Intelligence Initiative Act of 2020 (15
U.S.C. 9401).
<all>