[Congressional Bills 119th Congress]
[From the U.S. Government Publishing Office]
[S. 5057 Introduced in Senate (IS)]
<DOC>
119th CONGRESS
2d Session
S. 5057
To prohibit the use of certain artificial intelligence models across
the Federal Government, and for other purposes.
_______________________________________________________________________
IN THE SENATE OF THE UNITED STATES
July 21, 2026
Mr. Warner introduced the following bill; which was read twice and
referred to the Committee on Homeland Security and Governmental Affairs
_______________________________________________________________________
A BILL
To prohibit the use of certain artificial intelligence models across
the Federal Government, and for other purposes.
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 ``Safeguarding Against Fabricated
Exploitation Through Artificial Intelligence Act of 2026''.
SEC. 2. PROHIBITION ON THE USE OF CERTAIN ARTIFICIAL INTELLIGENCE
MODELS ACROSS THE FEDERAL GOVERNMENT.
(a) Definitions.--In this section:
(1) Administrator.--The term ``Administrator'' means the
Administrator of General Services.
(2) Appropriate congressional committees.--The term
``appropriate congressional committees'' means the Committee on
Commerce, Science, and Transportation of the Senate, the
Committee on the Judiciary of the Senate, the Committee on
Homeland Security and Governmental Affairs of the Senate, the
Committee on Oversight and Government Reform of the House of
Representatives, the Committee on Energy and Commerce of the
House of Representatives, and the Committee on the Judiciary of
the House of Representatives.
(3) Artificial intelligence model.--The term ``artificial
intelligence model'' means a capability or combined series of
capabilities that can generate realistic image or video outputs
for a given set of objectives, prompts, or input.
(4) Artificial intelligence risk management framework.--The
term ``Artificial Intelligence Risk Management Framework''
means the Artificial Intelligence Risk Management Framework, or
any successor document, and any associated guidance or
publications issued under section 22A of the National Institute
of Standards and Technology Act (15 U.S.C. 278h-1).
(5) Child pornography.--The term ``child pornography'' has
the meaning given the term in section 2256 of title 18, United
States Code.
(6) Child sexual abuse material.--The term ``child sexual
abuse material''--
(A) means child pornography; and
(B) an intimate visual depiction of an individual
who is less than 18 years of age.
(7) Covered application.--The term ``covered application''
means an artificial intelligence model that the Administrator
of General Services or the head of a Federal agency
determines--
(A) is not in compliance with--
(i) the standards document of the National
Institute of Standards and Technology numbered
``NIST AI 600-1'' and entitled ``Artificial
Intelligence Risk Management Framework:
Generative Artificial Intelligence Profile''
with respect to non-consensual intimate imagery
or child sexual abuse material within
``obscene, degrading, and/or abusive content'';
or
(ii) a comparably robust successor standard
or framework with respect to synthetic child
sexual abuse material or non-consensual
intimate images of adults;
(B) is subject to a determination by a Federal
court that the artificial intelligence model has
generated content depicting child pornography; or
(C) is subject to a determination by a Federal
court that the artificial intelligence model has
generated non-consensual intimate visual depictions of
an identifiable adult.
(8) Developer.--The term ``developer'' means a person that
develops an artificial intelligence model, including any person
that materially modifies and subsequently distributes an
artificial intelligence model.
(9) Federal agency.--The term ``Federal agency'' has the
meaning given the term ``agency'' in section 3502 of title 44,
United States Code.
(10) Intimate visual depiction.--The term ``intimate visual
depiction'' has the meaning given the term in section 1309 of
the Violence Against Women Act Reauthorization Act of 2022 (15
U.S.C. 6851).
(11) Technical control.--The term ``technical control''
means a technical control, or series of technical controls, to
prevent the ability of any publicly released version of an
artificial intelligence model to generate child pornography.
(b) Establishment of Performance Benchmarks.--
(1) Performance benchmark.--
(A) In general.--Not later than 90 days after the
date of enactment of this Act, the Director of the
National Institute of Standards and Technology shall
initiate a process to establish performance benchmarks,
consistent with the Artificial Intelligence Risk
Management Framework Generative Artificial Intelligence
Profile, to prevent the generation of obscene,
degrading, and abusive content, including synthetic
child sexual abuse material and non-consensual intimate
images of adults.
(B) Updates.--The Director of the National
Institute of Standards and Technology shall update, on
a periodic basis, the performance benchmarks
established under subparagraph (A) to address changes
in technology and circumvention practices.
(2) Testing program.--Not later than 180 days after the
date of the enactment of this Act, the Director of the National
Institute of Standards and Technology shall establish a
voluntary vendor test program consistent with the performance
benchmarks established under subparagraph (A).
(c) Prohibition on Federal Acquisition or Use of Covered
Applications.--
(1) In general.--The head of a Federal agency may not
procure a covered application or use such application on an
information technology system operated by such Federal agency
or by a contractor of such Federal agency.
(2) Implementation.--
(A) Initial removal.--Not later than 180 days after
the date of enactment of this Act, the head of each
Federal agency shall--
(i) remove any covered application from
each information technology system operated by
the Federal agency; and
(ii) ensure that each contractor of the
Federal agency removes any covered application
from each information technology system
operated by the contractor.
(B) Subsequent removals.--Not later than 180 days
after the date on which the Administrator or the head
of a Federal agency determines that an artificial
intelligence model constitutes a covered application,
including as a result of a review performed under
paragraph (3), the head of each Federal agency shall
carry out the requirements of clauses (i) and (ii) of
subparagraph (A) with respect to the covered
application.
(3) Continuous evaluation of available products and
services.--
(A) General services administration.--Effective on
the date that is 1 year after the date of enactment of
this Act, and not less frequently than every 90 days
thereafter, the Administrator shall review each product
or service offered within a contract vehicle or
purchasing program maintained by the General Services
Administration to determine whether the product or
service includes a covered application.
(B) Agency procurement review.--Effective on the
date that is 1 year after the date of enactment of this
Act, and not less frequently than annually thereafter,
the Chief Acquisition Officer or a senior procurement
executive of each Federal agency shall review each
product or service used by the Federal agency to
determine whether the product or service constitutes a
covered application.
(d) Safeguards.--
(1) In general.--If an artificial intelligence model that
is determined to constitute a covered application lacks a
vendor capable of making necessary technical modifications to
the artificial intelligence model in order to remove the status
of the artificial intelligence model as a covered application,
including for the purpose of subsection (e), the head of each
Federal agency may implement additional technical and
compliance safeguards, including those described in Reducing
Risks Posed by Synthetic Content (NIST AI 100-4) or any
successor publication, with respect to the artificial
intelligence model that effectively prevent the use of the an
artificial intelligence model for the generation of child
pornography or non-consensual intimate visual depictions of an
identifiable adult.
(2) Certification required.--With respect to the head of a
Federal agency who implements safeguards described in paragraph
(1) on a covered application, not later than 30 days after the
date of such implementation, the head of the Federal agency
shall issue a public certification that those safeguards are
sufficient to prevent the misuse of the covered application for
the generation of child pornography or non-consensual intimate
visual depictions of an identifiable adult.
(3) Congressional notification.--Not later than 7 days
after the date on which the head of a Federal agency issues a
certification under paragraph (2), the head of the Federal
agency shall submit to the appropriate congressional committees
a notification of the certification that identifies the
safeguards described in paragraph (1).
(e) Cure.--If the Administrator or the head of a Federal agency
determines that an artificial intelligence model constitutes a covered
application, the Administrator or the head of the Federal agency may
offer the vendor of the artificial intelligence model an opportunity to
modify the artificial intelligence model in order to remove the status
of the artificial intelligence model as a covered application at the
cost of the vendor to avoid the removal required under subsection
(c)(2).
(f) Prohibition; Private Right of Action.--
(1) Prohibition on products and services for
circumvention.--No person may deliberately manufacture, import,
or offer to the public a technology, product, service, device,
component, or part thereof that--
(A) is primarily designed or produced and promoted
for the purpose of circumventing, removing, or
tampering with a technical control;
(B) has only limited commercially significant or
expressive purpose or use other than to circumvent,
remove, or tamper with a technical control and is
promoted for a purpose described in subparagraph (A);
or
(C) is marketed by the person or another party
acting in concert with the person with the knowledge of
the person for use in circumventing, removing, or
tampering with a technical control.
(2) Private right of action.--
(A) In general.--The following persons may commence
a civil action in an appropriate district court of the
United States:
(i) An individual whose image or likeness
is contained in any child pornography.
(ii) An individual who has been injured by
a developer failing to implement sufficiently
robust technical controls to prevent their
image or likeness from being used to generate
child pornography, or been injured by any
person who has generated child pornography with
the image or likeness of that individual, in
violation of paragraph (1).
(iii) A developer whose technical controls
are subverted by a person in violation of
paragraph (1).
(B) Powers of the court.--In an action brought
under subparagraph (A), the court--
(i) may grant temporary and permanent
injunctions on such terms as it determines
reasonable to prevent or restrain a violation,
but in no event shall impose a prior restraint
on free speech or the press protected under the
First Amendment to the Constitution of the
United States;
(ii) at any time while an action is
pending, may order the impounding, on such
terms as it determines reasonable, of any
device or product that is in the custody or
control of the alleged violator and that the
court has reasonable cause to believe was
involved in a violation;
(iii) may award damages under subparagraph
(C);
(iv) in its discretion may allow the
recovery of costs by or against any party other
than the United States or an officer thereof;
(v) in its discretion may award reasonable
attorney's fees to the prevailing party; and
(vi) may, as part of a final judgment or
decree finding a violation, order the remedial
modification or the destruction of any device
or product involved in the violation that is in
the custody or control of the violator or has
been impounded under clause (ii).
(C) Award of damages.--
(i) In general.--Except as otherwise
provided in this section, a person committing a
violation of this section is liable for
either--
(I) the actual damages and any
additional profits of the violator, as
provided in clause (ii), or
(II) statutory damages, as provided
in clause (iii).
(ii) Actual damages.--In an action brought
under this paragraph, the court shall award to
the complaining party the actual damages
suffered by the party as a result of the
violation, and any profits of the violator that
are attributable to the violation and are not
taken into account in computing the actual
damages, if the complaining party elects such
damages at any time before final judgment is
entered.
(iii) Statutory damages.--At any time
before final judgment is entered in a case
brought under this paragraph, a complaining
party may elect to recover an award of
statutory damages of $50,000.
(D) Repeated violations.--In any action brought
under this paragraph in which the injured party
sustains the burden of proving, and the court finds,
that a person has violated this section within 3 years
after a final judgment was entered against the person
for another such violation, the court may increase the
award of damages up to triple the amount that would
otherwise be awarded, as the court considers just.
(E) Innocent violations.--In an action brought
under this paragraph, the court in its discretion may
reduce or remit the total award of damages in any case
in which the violator sustains the burden of proving,
and the court finds, that the violator was not aware
and had no reason to believe that its acts constituted
a violation.
<all>