Federal Agencies Continue Implementation of AI Executive Order

With Congress in summer recess and state legislative sessions waning, the Biden Administration continues to implement its October 2023 Executive Order on the Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence ("EO"). On July 26, the White House announced a series of federal agency actions under the EO for managing AI safety and security risks, hiring AI talent in the government workforce, promoting AI innovation, and advancing US global AI leadership. On the same day, the Department of Commerce released new guidance on AI red-team testing, secure AI software development, generative AI risk management, and a plan for promoting and developing global AI standards. These announcements-which the White House emphasized were on time within the 270-day deadline set by the EO-mark the latest in a series of federal agency activities to implement the EO.

AI Red-Teaming Testing and Risk Management

On July 26, NIST's U.S. AI Safety Institute ("US AISI") released initial public draft guidelines on "Managing Misuse Risk for Dual-Use Foundation Models." The guidelines outline practices for preventing malicious actors from recreating foundation models or deploying them to harm the public and include recommendations for developer red-teaming testing. US AISI is accepting public comments on the draft until September 9, 2024.

NIST released the final version of its Generative AI Profile and companion resource. The Generative AI Profile applies NIST's 2023 AI Risk Management Framework to 12 risks that are "unique to or exacerbated by" generative AI, including data privacy and information security, environmental impacts, harmful bias, harmful or obscene synthetic content, and IP risks.

These developments build on NIST's May 28, 2024, release of its Assessing Risks and Impacts of AI ("ARIA") pilot program, a test environment focusing on testing the risks and impacts of large language models, and the National Science Foundation's July 23, 2024, launch of an AI test beds initiative for studying AI methods and systems and January 24, 2024, launch of the National AI Research Resource ("NAIRR") pilot for compiling and sharing AI resources.

Secure AI Software Development Practices

NIST released the finalized version of its publication on "Secure Software Development Practices for Generative AI and Dual-Use Foundation Models," supplementing NIST's 2022 Secure Software Development Framework document. The new companion resource addresses risks related to malicious training data that can adversely affect the performance of generative AI systems. NIST also released Dioptra, an open-source software package designed to help AI developers and customers test the resiliency of AI models against adversarial attacks.

Global AI Standards

On July 26, the White House released the final version of its Implementation Roadmap for the May 2023 U.S. National Standards Strategy for Critical and Emerging Technology ("NSSCET"), following public comment on a draft roadmap released in June. That same day, NIST also released the final version of its "Plan for Global Engagement on AI Standards" ("Plan"), which incorporates stakeholder feedback and public comments on an earlier draft. Incorporating principles from the NIST AI Risk Management Framework and the US NSSCET, the Plan identifies over a dozen AI topic areas with clear or pressing needs for standardization, including defining key terminology for AI concepts, methods and metrics for assessing AI performance, risks, and benefits, and practices for maintaining and processing AI training data.

These recent federal agency actions are just a subset of ongoing activities to implement the Biden Administration's AI EO, and we anticipate more AI initiatives and developments as the White House approaches the one-year anniversary of the EO in October. These efforts build on a number of agency efforts in the first half of 2024, including a Commerce Department proposed rule regulating infrastructure-as-a-service providers and guidance on the use of AI by federal agencies from the White House Office of Management and Budget.

* * *

Follow our Global Policy Watch, Inside Global Tech, and Inside Privacy blogs for ongoing updates on key AI and other technology legislative and regulatory developments.

Public link

Please use the above public link if you want to share this noodl on another website.