S.Amdt. 3271

SA 3271. Mr. SCHUMER (for himself, Mr. Rounds, and Mr. Heinrich)
submitted an amendment intended to be proposed by him to the bill S.
4638, to authorize appropriations for fiscal year 2025 for military
activities of the Department of Defense, for military construction, and
for defense activities of the Department of Energy, to prescribe
military personnel strengths for such fiscal year, and for other
purposes; which was ordered to lie on the table; as follows:

At the appropriate place in title XVI, insert the
following:

SEC. ___. PHYSICAL AND CYBERSECURITY REQUIREMENTS FOR HIGHLY
CAPABLE ARTIFICIAL INTELLIGENCE SYSTEMS.

(a) Definitions.--In this section:
(1) Artificial intelligence.--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).
(2) Covered artificial intelligence technology.--The term
``covered artificial intelligence technology'' means a
technology specified in the guidance developed under
subsection (c)(3), including all components of that
technology, such as source code and numerical parameters of a
trained artificial intelligence system, and details of any
proprietary methods used to develop such a system.
(3) Covered entity.--The term ``covered entity'' means an
entity that enters into a Department of Defense contract that
engages in the development, deployment, storage, or
transportation of a covered artificial intelligence
technology.
(b) Findings.--Congress makes the following findings:
(1) Source code, numerical parameters, and related
technology associated with highly capable artificial
intelligence systems in the possession of private artificial
intelligence companies are an invaluable national resource
that would pose a grave threat to United States national
security if stolen by a foreign adversary through a cyber
operation or insider threat.

[[Page S6034]]

(2) Numerous foreign adversaries have the capacity to
engage in cyber operations to extract important data from
private companies, absent the most stringent cybersecurity
protections.
(c) Security Framework.--
(1) In general.--The Secretary of Defense, acting through
the Assistant Secretary of Defense for Cyber Policy, shall
develop a framework describing best practices for artificial
intelligence cybersecurity, physical security, and insider
threat mitigation to address or mitigate risks relating to
national security or foreign policy, including to protect
vital national resources from theft that would do grave
damage to the United States and to protect the proprietary
trade secrets used in the development of covered artificial
intelligence technologies which, if compromised, may create
risks to United States national security or foreign policy.
(2) Risk-based framework.--The framework developed under
paragraph (1) shall be risk-based, with stronger security
corresponding proportionally to the national security or
foreign policy risks posed by the artificial intelligence
technology being stolen or tampered with. The framework shall
include multiple security levels, where--
(A) at least one security level shall be equivalent to the
requirements described in NIST Special Publication 800-181
(relating to protecting controlled unclassified information
in nonfederal systems and organizations);
(B) at least one security level shall be equivalent to the
requirements described in NIST Special Publication 800-172
(relating to enhanced security requirements for protecting
controlled unclassified information); and
(C) at least one security level shall be stronger than NIST
Special Publication 800-172 (relating to enhanced security
requirements for protecting controlled unclassified
information) and shall describe a security posture capable of
mitigating risks posed by the highest threat actors,
including foreign intelligence agencies of peer and near-peer
nations.
(3) Covered artificial intelligence technologies.--
(A) Guidance.--The framework developed under paragraph (1)
shall provide clear guidance about which artificial
intelligence technologies are covered under the framework.
Such technologies shall be those that, if obtained by a
foreign adversary, would pose a grave threat to the national
security of the United States.
(B) Objective evaluation procedures.--Where feasible, the
guidance provided under subparagraph (A) shall be specified
in terms of objective evaluation procedures that measure or
estimate the national security implications of the artificial
intelligence technology, either before, during, or after it
has been developed.
(4) Use of existing frameworks.--To the maximum extent
feasible, the framework developed under paragraph (1) shall
be implemented using one or more existing cybersecurity
frameworks developed by the Department of Defense or other
Federal agencies, such as the Cybersecurity Maturity Model
Certification framework. Where needed, the Secretary may
augment those frameworks to implement additional security
levels as described in paragraph (2).
(d) Security Requirements.--
(1) In general.--The Secretary may amend the Defense
Federal Acquisition Regulation Supplement, or take other
similar action, to require covered entities to implement the
best practices described in the framework developed under
subsection (c).
(2) Risk-based rules.--Requirements implemented in rules
developed under paragraph (1) shall be as narrowly tailored
as practicable to the specific covered artificial
intelligence technologies developed, deployed, stored, or
transported by a covered entity, and shall be calibrated
accordingly to the different tasks involved in development,
deployment, storage, or transportation of components of those
covered artificial intelligence technologies.
(e) Reporting Requirements.--Not later than 180 days after
the date of the enactment of this Act, the Secretary, acting
through the Assistant Secretary, shall submit to the
congressional defense committees an update on the status of
implementation of the requirements of this section.
______