Sec. 6098. Vulnerability analysis study for artificial intelligence-enabled military applications
672 words·~3 min read·
/bill/118/s/2226/es/section-6098·A research copy — for the controlling text, always check the official state or federal source. Not legal advice.
Not later than one year after the date of the enactment of this Act, the Chief Digital and Artificial Intelligence Officer
(CDAO)of the Department of Defense shall complete a study analyzing the vulnerabilities to the privacy, security, and accuracy of, and capacity to assess, artificial intelligence-enabled military applications, as well as research and development needs for such applications. The study required by subsection
(a)shall cover the following: Research and development needs and transition pathways to advance explainable and interpretable artificial intelligence-enabled military applications, including the capability to assess the underlying algorithms and data models of such applications. Assessing the potential risks to the privacy, security, and accuracy of underlying architectures and algorithms of artificial intelligence-enabled military applications, including the following: Individual foundational artificial intelligence models, including the adequacy of existing testing, training, and auditing for such models to ensure models can be properly assessed over time. The interactions of multiple artificial intelligence-enabled military applications, and the ability to detect and assess new, complex, and emergent behavior amongst individual agents, as well as the collective impact, including how such changes may affect risk to privacy, security, and accuracy over time. The impact of increased agency in artificial intelligence-enabled military applications and how such increased agency may affect the ability to detect and assess new, complex, and emergent behavior, as well risks to the privacy, security, and accuracy of such applications over time. Assessing the survivability and traceability of decision support systems that are integrated with artificial intelligence-enabled military applications and used in a contested environment, including— potential benefits and risks to Department of Defense missions and operations of implementing such applications; and other technical or operational constraints to ensure such decision support systems that are integrated with artificial intelligence-enabled military applications are able to adhere to the Department of Defense Ethical Principles for Artificial Intelligence. Identification of existing artificial intelligence metrics, developmental, testing and audit capabilities, personnel, and infrastructure within the Department of Defense, including test and evaluation facilities, needed to enable ongoing identification and assessment under paragraphs
(1)through (3), and other factors such as— implications for deterrence systems based on systems warfare; and vulnerability to systems confrontation on the system and system-of-systems level. Identification of gaps or research needs to sufficiently respond to the elements outlined in this subsection that are not currently, or not sufficiently, funded within the Department of Defense. In carrying out the study required by subsection (a), the Chief Digital and Artificial Intelligence Officer shall coordinate with the following: The Director of the Defense Advanced Research Projects Agency (DARPA). The Under Secretary of Defense for Research and Evaluation. The Under Secretary of Defense for Policy. The Director for Operational Test and Evaluation (DOT&E) of the Department. As the Chief Digital and Artificial Intelligence Officer considers appropriate, the following: The Secretary of Energy. The Director of the National Institute of Standards and Technology. The Director of the National Science Foundation. The head of the National Artificial Intelligence Initiative Office of the Office of Science and Technology Policy. Members and representatives of industry. Members and representatives of academia. Not later than 180 days after the date of the enactment of this Act, the Chief Digital and Artificial Intelligence Officer shall provide the congressional defense committees a briefing on the interim findings of the Chief Digital and Artificial Intelligence Officer with respect to the study being conducted pursuant to subsection (a). Not later than one year after the date of the enactment of this Act, the Chief Digital and Artificial Intelligence Officer shall submit to the congressional defense committees a final report on the findings of the Chief Digital and Artificial Intelligence Officer with respect to the study conducted pursuant to subsection (a). The final report submitted pursuant to paragraph
(1)shall be submitted in unclassified for, but may include a classified annex. In this section, the term foundational artificial intelligence model means an adaptive generative model that is trained on a broad set of unlabeled data sets that can be used for different tasks, with minimal fine-tuning.