This is part of an ongoing series of Covington blogs on implementation of Executive Order 14028, “Improving the Nation’s Cybersecurity,” issued by President Biden on May 12, 2021 (the “Cyber EO”). The first blog summarized the Cyber EO’s key provisions and timelines, and subsequent blogs described the actions taken by various government agencies to implement the Cyber EO from June 2021 through June 2024. This blog describes key actions taken to implement the Cyber EO during July 2024. It also describes key actions taken during July 2024 to implement President Biden’s Executive Order on Artificial Intelligence (the “AI EO”), particularly its provisions that impact cybersecurity, national security, and software supply chain security.
NIST and NTIA Guidance Regarding Artificial Intelligence
The National Institute of Standards and Technology (“NIST”) released a cross-sectoral profile and companion resource to the AI risk management framework (“AI RMF”) that NIST released in January 2023. The profile implements the AI RMF functions and categories for Generative AI and offers insights into how risk can be managed across various stages of the AI lifecycle and for GAI as a technology.
The guidelines identify the following twelve areas of risks unique to or exacerbated by the development of Generative AI:
- Chemical, biological, radiological, or nuclear information or capabilities;
- Confabulation;
- Dangerous, violent, or hateful content;
- Data privacy;
- Environmental impacts;
- Harmful bias or homogenization;
- Human-AI configuration;
- Information integrity;
- Information security;
- Intellectual property;
- Obscene, degrading, and/or abusive content; and
- Value chain and component integration.
NIST encourages organizations to consider these categories of risks when attempting to identify and manage generative AI risks. NIST attempts to address these by offering proposed solutions to each category of risks.
Additionally, this month the NIST-run U.S. AI Safety Institute and the National Telecommunications and Information Administration (“NTIA”) both published new information on AI risks. The U.S. AI Safety Institute issued draft guidance on Managing Misuse Risk of Dual-Use Foundation Models. The draft guidance, which was developed in accordance with the AI EO, is focused on managing the risk that such models will be deliberately misused to cause harm. NTIA, as directed under the AI EO, released its Report on Dual-Use Foundation Models with Widely Available Model Weights. NTIA recommends that agencies develop new capabilities to monitor for potential risks, but refrain from immediately restricting the wide availability of open model weights in the largest AI systems.
Finally, NIST also released the final version of Special Publication 800-218A, “Secure Software Development Practices for Generative AI and Dual-Use Foundation Models” this month. The document fulfills a requirement of the AI EO – specifically, the requirement for NIST to “develop[] a companion resource to the Secure Software Development Framework to incorporate secure development practices for generative AI and for dual-use foundation models.” The publication recognizes that while there are significant similarities between generative AI/dual use foundational model software and traditional software, there are also unique risks that only apply to generative AI and dual use foundational model software. Notably, the publication recognizes that there may be risks associated with unknown datasets, tampering of model weights and functionality, and models that are too complex to allow for adequate inspection. The document describes practices that are specifically tailored to address these risks in the form of a “community profile” that supplements NIST SP 800-218, which addresses the risks of software vulnerabilities in the development context more generally. The community profile is meant to be tailored and adopted based on the specific needs of the developer, and includes various “practices” and “tasks” that should be considered by the organization. Ultimately, the practices and tasks described by this publication could be adopted by the Government as an extension of various existing initiatives, such as the requirement for attestation to secure development practices or a third party assessment in order to develop software used by an agency, as described here.
Executive Office of the President Releases Memorandum Regarding Administration Cybersecurity Priorities for the FY 2026 Budget
The Executive Office of the President released a memorandum on July 10, outlining the Biden Administration’s cross-agency cybersecurity investment priorities for formulating its fiscal year (“FY”) 2026 budget proposal. The memorandum details the priorities agencies should focus on in their budget submissions and outlines the process by which the Office of Management and Budget (“OMB”) and the Office of the National Cyber Director will review agency budget submissions to ensure they adequately address and are consistent with overall cybersecurity strategy and policy, aiding agencies’ multi-year planning through the regular budget process.
The memorandum notes that agencies budget submissions should focus on increasing the maturity of information systems consistent with the National Cybersecurity Strategy’s five pillars to enhance national security: (1) Defend Critical Infrastructure, (2) Disrupt and Dismantle Threat Actors, (3) Shape Market Forces to Drive Security and Resilience, (4) Invest in a Resilient Future, and (5) Forge International Partnerships to Pursue Shared Goals. In the accordance with the Biden Administration’s cybersecurity priorities, among the areas of emphasis for agencies are the following:
- Scaling public-private collaboration to defend critical infrastructure;
- Implementing zero trust architecture;
- Improving basic cybersecurity requirements, including by only using software that is provided by producers who can attest to compliance with Government-specified secure software development practices; and
- Increasing secure use and maintenance of open source software into IT and governance structures.
The memorandum serves to further emphasize the administration’s interest in improving agencies’ basic cybersecurity requirements and gives contractors a view into the priorities of agencies going forward.
CISA Updates Software Bill of Materials Frequently Asked Questions
On July 16, the Cybersecurity and Infrastructure Agency (“CISA”) announced via X that it updated its Software Bill of Materials (“SBOM”) Frequently Asked Questions (“FAQ”). This is the latest step CISA has taken to advance the adoption of SBOMs in federal procurements. The updated SBOM FAQ provides general introductory information on SBOMs, common use cases for SBOMs, their benefits, common misconceptions and concerns, the processes for creating, distributing, and sharing an SBOM. The FAQ could be helpful to contractors seeking more information about SBOMs, particularly as they seek to comply with forthcoming rules. This includes the proposed FAR rule that may require contractors to develop and maintain a SBOM for any software used in the performance of a government contract, and the Office of Management and Budget’s secure software development requirements, which indicates that SBOMs may be required if attestations to secure software development requirement implementation are not sufficient.
OMB Issues Memorandum on Modernizing the Federal Risk and Authorization Management Program
OMB released a memo on July 25 outlining substantial changes to the Federal Risk and Authorization Management Program (“FedRAMP”). This memo was issued in connection with the FedRAMP Authorization Act and replaces the December 2011 memo establishing FedRAMP as the government risk assessment program for cloud service providers. It also follows a draft that the Office of Management and Budget issued in October 2023 for public comment. The memo outlines OMB’s plan to strengthen FedRAMP’s approach to security, increase procurement of cloud products and services, and broaden FedRAMP’s reach.
The memo outlines certain changes to FedRAMP. Of note to contractors, the memo implements several changes to the FedRAMP authorization process (including the creation of an additional “program level” path to authorization); clarifies FedRAMP’s scope to include “cloud computing products and services (such as [Infrastructure-as-a-Service] (IaaS), Platform-as-a-Service (PaaS), and SaaS) that create, collect, process, store, or maintain Federal information on behalf of a Federal agency, and that are not otherwise specified as out of scope;” and encourages the adoption of a standardized continuous monitoring framework.
Within 180 days of its release, agencies are expected to issue policies aligned with the memo. Within one year of the memo’s release, GSA will issue a plan to structure FedRAMP “to encourage the transition of Federal agencies away from the use of Government-specific cloud infrastructure.”
FCC Adopts Order Establishing Voluntary IoT Labeling Program
On July 30, the Federal Communications Commission (“FCC”) adopted an order creating a voluntary cybersecurity labeling program for Internet of Things (“IoT”) devices. The FCC’s new order relates to Section 4(t) of the Cyber EO, which directs certain agencies to identify IoT cybersecurity criteria for a consumer labeling program. The Inside Global Tech blog has full coverage of this order.
Senate Homeland Security Committee Approves AI Procurement Reform Bill
On July 31, the Senate Homeland Security Committee cleared for floor consideration the PREPARED for AI Act. The bill would codify guidelines issued in March by the OMB on implementing President Biden’s AI EO. The bill, sponsored by Senators Gary Peters and Thom Tillis, would create a Chief Artificial Intelligence Officers Council that would coordinate the development and use of AI among federal agencies and form agency-specific AI governance boards to manage and oversee AI activities. If the bill passed, agencies would establish a risk-classification system to identify high-risk AI. Contractors engaging in high-risk uses of AI would be subject to ongoing monitoring and reporting obligations and would begin to see contract terms governing AI capabilities, should the bill pass, and would generally supplement the required actions under Biden’s AI EO.