Application of artificial intelligence policies of the intelligence community to publicly available models used for intelligence purposes

(a) In general Section 6702 of the Intelligence Authorization Act for Fiscal Year 2023 (50 U.S.C. 3334m) is amended— (1) by redesignating subsection (c) as subsection (e); (2) by inserting after subsection (b) the following: (c) Application of policies to publicly available models used for intelligence purposes In carrying out subsections (a) and (b), the Director shall ensure that the policies established under such subsections apply to the greatest extent possible to artificial intelligence models generally available to the public in any context in which they are used for an intelligence purpose and hosted in classified environments. (d) Common testing standards and benchmarks (1) Establishment The Chief Artificial Intelligence Officer of the Intelligence Community, or any provider of common concern designated by the Director of National Intelligence, shall establish standards for testing of artificial intelligence models, including common benchmarks and methodologies for the performance of artificial intelligence models across common use cases, including targeting, machine translation, object detection, and object recognition. Benchmarks and methodologies shall establish higher performance standards for any high-impact artificial intelligence use case, including any artificial intelligence system task whose output (directly or indirectly) could serve as an input for a lethal application. (2) Identification of computing model The Chief Artificial Intelligence Officer of the Intelligence Community shall convene the Intelligence Community Chief Artificial Intelligence Officer Council to identify an appropriate computing environment, at a level (or multiple levels) of classification deemed appropriate, for elements of the intelligence community to engage in testing and evaluation of models prior to acquisition.; and (3) by adding at the end the following: (f) Limitation Under the policies established pursuant to subsection (a)(1), no office or employee of the intelligence community may direct or pressure a vendor or prospective vendor to alter a model to favor a particular viewpoint in a manner that would limit its ability to serve as a neutral, nonpartisan tool that prioritizes accuracy. (g) Definitions (1) Intelligence purpose defined In this section, the term intelligence purpose means the collection, analysis, or other mission-related intelligence activity. (2) Guidance regarding definitions of high-impact artificial intelligence Not later than 30 days after the date of the enactment of this subsection, the Director of National Intelligence shall issue guidance to the heads of elements of the intelligence community to ensure consistency and accuracy in each element’s interpretation of the definition of high-impact artificial intelligence systems and high-impact artificial intelligence use cases to apply to each element’s respective missions.. (b) Updates The Director shall make such revisions to Intelligence Community Directive 505 (relating to Artificial Intelligence) and other relevant documents as the Director considers necessary to ensure compliance with subsection (c) of section 6702 of such Act, as added by subsection (a).