Latest White House Memo Adds to Growing Federal AI Policy Framework

The federal government is keen to expand the policies and guidance around AI technologies. Following the release of the AI EO nearly a year ago, the White House has issued several memos on the federal use and acquisition of AI. The latest memorandum titled "Advancing the United States’ Leadership in Artificial Intelligence; Harnessing Artificial Intelligence to Fulfill National Security Objectives; and Fostering the Safety, Security, and Trustworthiness of Artificial Intelligence" addresses the incorporation of AI in national security systems, an area which was not typically covered by previous AI policies.

Specifically, the latest memo, “directs actions to strengthen and protect the United States AI ecosystem; improve the safety, security, and trustworthiness of AI systems developed and used in the United States; enhance the United States Government’s appropriate, responsible, and effective adoption of AI in service of the national security mission; and minimize the misuse of AI worldwide.”

The memorandum seeks to achieve three objectives:

• The U.S. as a global leader in the development of safe, secure and trustworthy AI
• The U.S. harnessing AI with appropriate safeguards to meet national security missions
• The U.S. developing a framework to advance international AI governance

To meet the first objective, the memorandum dedicates a section to promoting and securing U.S. foundational AI capabilities. Requirements in the section include streamlining the process to attract and accelerate the hire and move of foreign individuals to the U.S. with technical expertise to improve U.S. AI competitiveness and conducting assessments of the U.S. private sector’s AI competitive edge. Moreover, the memo calls on developing advanced AI-dedicated computational infrastructure, including using the National AI Research Resource (NAIRR) to distribute computational resources and providing incentives for the construction of AI-enabling infrastructure.

The memorandum also expands authorities for the AI Safety Institute (AISI) at Commerce/NIST to help manage risks to AI safety and security. For instance, Commerce is charged with establishing a capability to lead voluntary unclassified pre-deployment safety testing for frontier AI models on behalf of the federal government. The memo also charges AISI with issuing guidance for AI developers on how to test, evaluate and manage risks arising from dual-use foundational models. Additionally, AISI will recommend metrics for assessment AI systems’ capabilities and limitations in categories such as science, mathematics, code generation and general reasoning.

In terms of managing risks within classified systems, the memo charges the NSA, in conjunction with AISI, to develop the capability to perform rapid systematic classified testing of AI models’ capacity to detect cyber threats. The memo also requires the NNSA, in conjunction with AISI and NSA, to do the same for AI models in relation to nuclear and radiological risks. DHS is charged with working with the aforementioned agencies for chemical and biological threats.

With regards to the second objective of responsibly harnessing AI for national security missions, the memorandum requires an updated set of government-wide procedures for attacking and retaining AI and AI-enabling talent for national security, while also charging IC entities to identify education and training opportunity to increase AI competencies within the current workforce.

The memo also calls for the coordination and acceleration of AI acquisition and procurements. To do so, DOD and ODNI are to establish a working group to address the issues of AI procurement for use on National Security Systems. Thereafter, guidance is go be developed by DOJ for constitutional considerations raised by the IC’s acquisition and use of AI.

For the third objective, to foster an international AI governance landscape, the memo calls on State and other agencies to, “produce a strategy for the advancement of international AI governance norms in line with safe, secure, and trustworthy AI, and democratic values, including human rights, civil rights, civil liberties, and privacy.  This strategy shall cover bilateral and multilateral engagement and relations with allies and partners.  It shall also include guidance on engaging with competitors, and it shall outline an approach to working in international institutions.”