Requires the Director of National Intelligence to implement policies for the third-party acquisition, development, and use of artificial intelligence capabilities.
Instructs the Director of National Intelligence, in consultation with the Director of the Office of Management and Budget and other intelligence officials to maintain, develop, and annually revise various AI policies.
Requires the establishment of a lexicon relating to machine learning and AI use by the intelligence community.
Requires the establishment of guidelines for evaluating model performance by specifying conditions for AI capability monitoring, documenting performance objectives, specifying model auditing processes, and specifying conditions for vulnerability testing and privacy evaluations.
Requires the establishment of guidelines for tracking dependencies in related systems, documentation requirements for capabilities procured from third-party and training data usage, and guidance on the usage of third-party-trained models.
Key facts
🏛️ This document has been enacted by the United States Congress.
For authoritative text and metadata, visit the official source.
🎯 This document primarily applies to the government, rather than the private sector.
📜 This document's name is National Defense Authorization Act for Fiscal Year 2024, Section 7513 ("Policies established by Director of National Intelligence for artificial intelligence capabilities").
AGORA also tracks this document under the name FY2024 NDAA, Section 7513 ("Policies established by Director of National Intelligence for artificial intelligence capabilities"). It is part of FY2024 NDAA.
↳ This document is part of a longer one: FY2024 NDAA.
Some AGORA documents are "split off" from longer documents that mix AI
and non-AI content, such as omnibus authorization or appropriations laws
in the United States Congress. Read more >>
Themes
AI risks, applications, governance strategies, and other themes addressed in AGORA documents.
This tag applies to documents that address whether humans can understand (a) the mechanisms underlying an AI system’s operation and/or (b) the meaning of the systems’ output in the context of use.
This tag applies to documents that address the use or presence of personally identifying information in an AI system's inputs or outputs, and any related protections or mitigations.
This tag applies to documents that govern the security of digital systems associated with AI, including systems involved in development and training, systems housing related intellectual property, and computing infrastructure for deployed models.
This tag applies to documents that facilitate, require, set conditions on, or otherwise address the convening of different stakeholders in AI systems - for example, to share feedback or to participate in its development or deployment.
This tag applies to documents that address disclosure of whether and how an AI system is deployed; for example, requiring users of government services to be notified when AI systems are used to determine how the services are provided.
This tag applies to documents that address the form of disclosure of information about AI systems, such that information is required or encouraged to be disclosed in a prescribed, standardized form intended to make the information easily accessible and understandable.
This tag applies to documents that discuss AI evaluations by a disinterested counterparty or third party, such as a customer or a professional auditing firm.
This tag applies to documents that discuss evaluations that assess whether deployed AI systems meet the expectations specified or claimed prior to deployment.
This tag applies to documents that discuss AI evaluations by a disinterested counterparty or third party, such as a customer or a professional auditing firm.
This tag applies to documents that support, encourage, or require the development of, or impose conditions on, other, subsequent public or private governance instruments related to the development or deployment of AI systems.
This tag applies to documents that address government support for AI-related research and development (R&D). R&D includes activities meant to create or improve technologies, products, and scientific knowledge.
This tag applies to documents that address government support for AI-related research and development (R&D). R&D includes activities meant to create or improve technologies, products, and scientific knowledge.
Full text
This is an unofficial copy. The document has been
archived and reformatted in plaintext for AGORA. Footnotes, tables, and
similar material may be omitted. For the official text, visit the original source.
SEC. 7513. POLICIES ESTABLISHED BY DIRECTOR OF NATIONAL INTELLIGENCE FOR ARTIFICIAL INTELLIGENCE CAPABILITIES.
(a) In General.--Section 6702 of the Intelligence Authorization Act for Fiscal Year 2023 (50 U.S.C. 3334m) is amended--
(1) in subsection (a), in the matter preceding paragraph (1), by striking subsection (b)'' and inserting subsection (c)'';
(2) by redesignating subsection (b) as subsection (c); and
(3) by inserting after subsection (a) the following:
Amends Intelligence Authorization Act to update subsections and insert new provisions for AI capabilities policies.
Amends Intelligence Authorization Act to update subsections and insert new provisions for AI capabilities policies.
``(b) Policies.--
``(1) In general.--In carrying out subsection (a)(1), not later than 1 year after the date of the enactment of the Intelligence Authorization Act for Fiscal Year 2024, the Director of National Intelligence, in consultation with the heads of the elements of the intelligence community, the Director of the Office of Management and Budget, and such other officials as the Director of National Intelligence determines appropriate, shall establish the policies described in paragraph (2).
``(2) Policies described.--The policies described in this paragraph are policies for the acquisition, adoption, development, use, coordination, and maintenance of artificial intelligence capabilities that--
``(A) establish a lexicon relating to the use of machine learning and artificial intelligence developed or acquired by elements of the intelligence community;
Establishes policies for AI capabilities' acquisition, adoption, use, coordination, and maintenance within the intelligence community.
Establishes policies for AI capabilities' acquisition, adoption, use, coordination, and maintenance within the intelligence community.
``(B) establish minimum guidelines for evaluating the performance of models developed or acquired by elements of the intelligence community, such as by--
``(i) specifying conditions for the continuous monitoring of artificial intelligence capabilities for performance, including the conditions for retraining or retiring models based on performance;
``(ii) documenting performance objectives, including specifying how performance objectives shall be developed and contractually enforced for capabilities procured from third parties;
``(iii) specifying the manner in which models should be audited, as necessary, including the types of documentation that should be provided to any auditor; and
``(iv) specifying conditions under which models used by elements of the intelligence community should be subject to testing and evaluation for vulnerabilities to techniques meant to undermine the availability, integrity, or privacy of an artificial intelligence capability;
Establishes guidelines for evaluating AI models' performance, monitoring, objectives, auditing, and vulnerability testing.
Establishes guidelines for evaluating AI models' performance, monitoring, objectives, auditing, and vulnerability testing.
``(C) establish minimum guidelines for tracking dependencies in adjacent systems, capabilities, or processes impacted by the retraining or sunsetting of any model described in subparagraph (B);
``(D) establish minimum documentation requirements for capabilities procured from third parties, aligning such requirements, as necessary, with existing documentation requirements applicable to capabilities developed by elements of the intelligence community;
``(E) establish minimum standards for the documentation of imputed, augmented, or synthetic data used to train any model developed, procured, or used by an element of the intelligence community; and
``(F) provide guidance on the acquisition and usage of models that have previously been trained by a third party for subsequent modification and usage by such an element.
``(3) Policy review and revision.--The Director of National Intelligence shall annually review or revise each policy established under paragraph (1).''.
(b) Conforming Amendment.--Section 6712(b)(1) of such Act (50 U.S.C. 3024 note) is amended by striking section 6702(b)'' and inserting section 6702(c)''.
Establishes guidelines, documentation standards, and provides guidance for AI models within the intelligence community.
Establishes guidelines, documentation standards, and provides guidance for AI models within the intelligence community.