Learning Objectives

After this lecture, L5 leaders will be able to: (1) differentiate the legal and operational forms of AI advisory structures, statutorily-chartered committees under the Federal Advisory Committee Act (FACA, 5 U.S.C. App. 2), agency AI Governance Boards required by OMB M-24-10, Chief AI Officer councils, cross-agency coordination bodies like the CIO Council's AI Community of Practice and the Interagency AI Council, and mission-specific bodies such as the National AI Advisory Committee (NAIAC) chartered under Section 5104 of the National AI Initiative Act of 2020; (2) charter an agency AI Governance Board under M-24-10 with the right membership (Chief AI Officer as chair, General Counsel, CIO, CISO, Chief Privacy Officer, civil-rights or equity lead, mission owner, Inspector General as observer), the right decision rights, and the right relationship to procurement, privacy, security, and program authorities; (3) design advisory committees that produce actionable output rather than bureaucratic debris by setting clear questions, defined deliverables, sunset dates, and public-reporting obligations; (4) navigate FACA compliance for advisory committees that include non-federal members, charter, designated federal officer (DFO), public meeting and minutes obligations, balanced membership, GSA oversight, and biennial renewal, and know when to invoke FACA exemptions narrowly; (5) integrate state and local AI advisory mechanisms with federal structures, including governor's AI task forces, state CIO associations, the NASCIO AI community, and municipal surveillance-ordinance review boards; and (6) measure advisory-structure effectiveness against concrete outputs (decisions supported, policies promulgated, audits completed, harms avoided) rather than input metrics (meetings held, members appointed).

Key Topics Covered

This lecture covers seven concrete structures that L5 leaders will either build, run, or serve on. First, the agency AI Governance Board required by OMB M-24-10 Section 3(b). Second, the Chief AI Officer Council chartered in the OMB Chief AI Officer Guidance (2024) as the CFO Act interagency body. Third, the National AI Advisory Committee (NAIAC) chartered under the National AI Initiative Act of 2020, reporting to the President and the National AI Initiative Office hosted at OSTP, with a Law Enforcement Subcommittee. Fourth, the Federal CIO Council's AI Community of Practice, which functions as a practitioner network rather than a decision body. Fifth, mission-specific advisory structures: the Defense Innovation Board AI ethics work that produced the 2020 DoD AI Ethical Principles, the Department of Energy's AI Advancement Council, HHS's AI Council, and the VA's National AI Institute advisory structures. Sixth, state and local structures: New York City's Algorithmic Management Policy, California's GenAI state initiative governance, Texas's AI Advisory Council created by HB 2060 (2023), and Seattle's Surveillance Advisory Working Group. Seventh, civil-society and academic bodies that function as de facto advisory: the National Academies' CSTB AI committees, Stanford HAI's policy practice, Brookings AI Equity Lab, AI Now, the Center for Security and Emerging Technology (CSET) at Georgetown, and the Ada Lovelace Institute on the international side. We will also treat the legal scaffolding: FACA, the Federal Register Act notice requirements, sunshine law equivalents at state level, and the ethics statutes (18 U.S.C. § 208) that govern conflicts of interest for special government employees and committee members.

Why This Matters for Government

AI advisory structures are the organizational machinery through which federal agencies, state governments, cities, tribal governments, and international bodies turn AI policy into operational decisions. Most AI governance failures in US government are not failures of policy text; they are failures of the structures that were supposed to interpret, apply, and enforce the policy. The NIST AI Risk Management Framework 1.0 (January 2023) provides a functional framework. OMB Memorandum M-24-10 (March 2024) sets minimum practices and requires every CFO Act agency to designate a Chief AI Officer and stand up an AI Governance Board. Executive Order 14110 (October 2023) sets direction. The National AI Initiative Act of 2020 created the National AI Advisory Committee (NAIAC). None of these documents makes a decision. They define who must decide and how. The advisory structure is the decision engine, and structure determines which people are in the room when the agency chooses to deploy, waive, pause, or retire an AI system that affects citizens.

The case record demonstrates why structure is decisive. When Michigan deployed MIDAS between 2013 and 2015, the failure that produced roughly forty thousand false fraud accusations and more than twenty million dollars in settlements was not primarily the absence of rules; it was the absence of a governance board that would have refused to approve minimum-review automated fraud determinations. When the IRS paused its ID.me facial-verification rollout in February 2022 after public outcry and letters from Senators Wyden, Warren, and Menendez, the root cause was not the technology but the fact that no advisory structure with civil-rights, privacy, Section 508 accessibility, and tax-administration authority had been constructively engaged before the rollout. When the Department of Veterans Affairs integrated AI into clinical decision support through the National Artificial Intelligence Institute (NAII), the successful integration was driven by a disciplined advisory process with clinicians, ethicists, veterans' service organizations, and academic affiliate medical schools. When the Centers for Medicare and Medicaid Services considered AI-assisted prior authorization, the review structure included CMS leadership, HHS Office of Inspector General, patient advocates, and the Quality Improvement Organization network, and the resulting April 2024 rule CMS-0057-F reflects that review. The Dutch childcare benefits (toeslagenaffaire) scandal and SyRI in the Netherlands, the UK Post Office Horizon matter, Houston Federation of Teachers v. Houston ISD (S.D. Tex. 2017), and Compas (State v. Loomis, Wisconsin 2016) all trace their worst harms to missing or captured advisory structures rather than to technological inadequacy alone. Structure matters because people matter, and structure determines whose expertise, interests, and lived experience inform the decision.

Federal advisory structures in the United States fall into four legally distinct categories that carry different obligations. First, Federal Advisory Committee Act (FACA) chartered committees: any group established or utilized by the executive branch to obtain advice that includes non-federal members is generally subject to FACA and therefore requires a charter, a Designated Federal Officer, open meetings with limited Sunshine Act exceptions, published minutes, balanced membership, GSA Committee Management Secretariat oversight, and biennial renewal. NAIAC is FACA-chartered and operates with a Law Enforcement Subcommittee and other working groups; the Defense Innovation Board, the DOE FACA committees that review AI in National Labs, the DHS Homeland Security Advisory Council, the HHS Advisory Committee on Human Research Protections, and the NTIA and NIST external advisory structures are other federal examples. Second, intragovernmental bodies such as the Chief AI Officer Council convened by OMB: these are composed entirely of federal employees, are not FACA-subject, but remain subject to federal records laws (44 USC Chapter 29) and the Presidential and Federal Records Act Amendments. Third, agency governance boards required by M-24-10 Section 3(b): these are internal decision bodies chartered by agency order, typically chaired by the CAIO, with membership including the General Counsel, CIO, CISO, Chief Privacy Officer (the SAOP under OMB Circular A-130), civil rights officer, mission owners, senior procurement executive, and Inspector General liaison as observer. Fourth, de facto advisory bodies including civil society coalitions, academic panels, and industry groups whose recommendations agencies consider but which are not formally 'utilized' under FACA; these are not regulated but still create accountability expectations and, if mismanaged, have triggered FACA litigation (e.g., Public Citizen v. NRC, Cheney v. US District Court 2004).

Advisory structure legitimacy is built on three pillars. First, composition breadth: a committee that includes only industry voices will be seen, correctly, as captured; one that includes only critics will be seen as symbolic; credible committees include industry practitioners, academic researchers, civil society advocates, mission operators, and affected community representatives. Second, conflicts management under 18 USC 208 and 5 CFR 2640: Special Government Employees file OGE Form 450 confidential financial disclosures, recuse from matters affecting personal financial interests, and operate under institutional conflict rules where employer relationships create bias even without direct financial interest. Third, durability: appointments that lapse without renewal pipelines lose institutional memory at exactly the moment the committee matures. The CAIO Council, the CIO Council AI Community of Practice, NAIAC, OSTP's National AI Initiative Office, and the GSA AI Center of Excellence together form an interlocking federal advisory network; L5 leaders should map their agency's relationship to each and route the appropriate question to the appropriate body. State examples include Texas HB 2060 (2023) AI Advisory Council, New York City's Algorithmic Management Policy and Local Law 35, California's GenAI executive order N-12-23 (2023), Washington State RCW 43.386, Seattle's Surveillance Advisory Working Group under SMC 14.18, and tribal bodies emerging in the Navajo Nation and the Confederated Tribes. International bodies including the EU AI Board under the EU AI Act, the UK AI Safety Institute, Singapore's Model AI Governance Framework advisory work, and OECD AI committees provide comparative structure for US agencies operating internationally. The remainder of this seminar walks through how to charter, run, staff, and sunset each kind of body, anchored to case studies from the VA, CMS, DHS, DOD, IRS, NIST, and peer jurisdictions.

Advisory Structures Are Where Policy Meets Operations

Most AI governance failures in government are not failures of policy text. They are failures of the structures that were supposed to interpret, apply, and enforce the policy. The NIST AI RMF provides a functional framework. OMB M-24-10 sets minimum practices. EO 14110 sets direction. None of these documents makes a decision; they define who must decide and how. The advisory structure is the decision engine. When Michigan deployed MiDAS, the failure was not the absence of rules. It was the absence of a governance board that would have refused to approve minimum-review fraud determinations. When the IRS paused its ID.me facial-verification rollout in February 2022 after public outcry, the root cause was not the technology; it was that no advisory structure with civil-rights, privacy, and tax-administration authority had been constructively engaged before the rollout. When the Department of Veterans Affairs integrated AI into clinical decision support, the successful integration was driven by the VA's National AI Institute's use of a disciplined advisory process with clinicians, ethicists, and veterans' service organizations. When the Centers for Medicare & Medicaid Services considered AI-assisted prior authorization, the review structure included CMS, HHS OIG, patient advocates, and the CMS Quality Improvement Organization network, and the resulting April 2024 rule (CMS-0057-F) reflects that review. Structure matters because people matter, and structure determines which people are in the room.

The Legal Taxonomy Leaders Must Understand

Federal advisory structures in the United States fall into four legally distinct categories that carry different obligations. (1) FACA-chartered advisory committees: any group established or utilized by the executive branch to obtain advice or recommendations that includes non-federal members is generally subject to FACA. FACA requires a charter, a Designated Federal Officer (DFO), open meetings (with limited closed-meeting exceptions under the Government in the Sunshine Act and the Privacy Act), published minutes, balanced membership, GSA oversight through the Committee Management Secretariat, and biennial renewal. NAIAC is FACA-chartered. (2) Intragovernmental bodies: councils composed entirely of federal employees (such as the Chief AI Officer Council) are not FACA-subject but remain subject to federal records laws (44 U.S.C. Chapter 29), the Federal Advisory Committee Act's 'utilization' rules if they begin to consult non-federal stakeholders systematically, and internal ethics rules. (3) Agency governance boards: the M-24-10-required AI Governance Boards are agency-internal governance bodies, typically chartered by agency order, with decision authority rather than advisory-only scope. (4) De facto advisory bodies: civil-society coalitions, academic panels, and industry bodies whose recommendations the government considers but which are not 'utilized' in the FACA sense; these are not formally regulated but do create accountability expectations. The legal distinctions matter because getting them wrong, for example, running a de facto FACA committee without chartering it, has triggered litigation in the past (e.g., the Public Citizen v. NRC line of cases and the Cheney v. U.S. District Court 2004 energy task force litigation). L5 leaders must know which box they are in.

Charter a Governance Board That Actually Works

Under OMB M-24-10, every CFO Act agency must stand up an AI Governance Board, chaired by the Chief AI Officer. The membership composition determines the board's capacity. A working board has: the CAO as chair, the General Counsel or senior delegate, the CIO, the CISO, the Chief Privacy Officer (the SAOP under OMB Circular A-130), the civil-rights officer or equity lead, the mission owner relevant to each agenda item (rotated), the senior procurement executive for contract items, and the Inspector General liaison as observer. Less functional boards have: an IT-only composition that treats AI as an engineering question; a policy-only composition that never sees operational reality; or a rotating cast that cannot build institutional memory. The decision rights should include: approval of new AI use cases classified as safety- or rights-impacting; approval of waivers under M-24-10 (with notice to OMB); approval of impact assessments and monitoring plans; review of incidents and corrective actions; oversight of the annual AI Use Case Inventory submission; and review of procurement items over defined thresholds. The cadence should balance urgency (emergency items reviewable within business days) with deliberation (standing monthly meetings with a quarterly strategic review). Minutes should be kept, decisions memorialized, and an annual public summary published alongside the Use Case Inventory. A board that meets rarely, decides little, and produces no public record is a liability in an IG audit.

FACA Advisory Committees for External Input

When an agency needs structured input from non-federal experts, industry, academia, civil society, subject-matter specialists, FACA is typically the right vehicle. NAIAC is the prototype at the top of the federal government. Within agencies, examples include the Department of Commerce's NTIA/NIST advisory structures, the HHS Advisory Committee on Human Research Protections, the DOE FACA committees that review AI in the National Labs, and the DHS Homeland Security Advisory Council, which has had AI-related subcommittees. Chartering a FACA committee is a discipline. The charter articulates purpose, scope, membership, reporting, and sunset. GSA's Committee Management Secretariat reviews. The DFO manages meetings, notices, minutes, and member ethics briefings; 18 U.S.C. § 208 and the regulations for Special Government Employees apply. Balanced membership is not optional. It is statutory. Meetings must be announced in the Federal Register at least 15 calendar days in advance (with narrow exceptions). Closed-meeting requests must be supported under the Sunshine Act exemptions. Products must be published. A committee that does not publish never produces durable value. At the state level, equivalent sunshine laws (California Bagley-Keene, Texas Open Meetings Act, Illinois Open Meetings Act, New York Open Meetings Law) impose similar obligations. L5 leaders should assume that advisory work is public work and design accordingly.

Composition, Conflicts, and Legitimacy

Advisory-structure legitimacy is built on three pillars. First, composition breadth: a committee that includes only industry voices will be seen, correctly, as captured; one that includes only critics will be seen as symbolic. Credible committees include industry practitioners, academic researchers, civil-society advocates, mission operators, and affected community representatives. Second, conflicts management: Special Government Employees must file OGE Form 450 (confidential financial disclosures), recuse from matters affecting personal financial interests, and operate under 18 U.S.C. § 208. Agencies should also manage 'institutional' conflicts where an employer relationship creates bias even without direct financial interest. Third, durability: appointing a member for a two-year term without a renewal or replacement pipeline means the committee loses institutional memory at exactly the moment it matures. The Biden administration's NAIAC (established 2022) is a useful study: membership was deliberately diversified across private-sector, academic, civil-society, and practitioner voices; subcommittees including the Law Enforcement Subcommittee produced actionable reports; and the charter was renewed under its biennial requirement. Study the NAIAC charter and the membership list as a benchmark before you charter your own body.

Cross-Agency Coordination and the Chief AI Officer Council

Advisory coordination across agencies is handled by several federal bodies. The Chief AI Officer Council, chartered under the OMB Chief AI Officer guidance, is the CFO Act interagency body. It produces shared templates, harmonizes impact-assessment practice, coordinates M-24-10 implementation across agencies, and provides an interface with OMB's Office of the Federal CIO. The CIO Council's AI Community of Practice (AI CoP) is a practitioner-level network where the people doing the day-to-day work of M-24-10 implementation at agencies share patterns, failures, and tooling. The National AI Initiative Office at OSTP coordinates federal AI policy more broadly under the National AI Initiative Act. The GSA AI Center of Excellence provides implementation-level technical advisory services to agencies. Together these bodies form a network: the CAO Council makes policy-aligned decisions, the AI CoP shares practitioner knowledge, NAIAC provides external advice, the NAIIO coordinates at White House level, and the GSA CoE provides hands-on help. L5 leaders should map their agency's relationship to each and use the appropriate body for each kind of question: policy interpretation to CAO Council, shared tooling to AI CoP, implementation help to GSA CoE, external validation to NAIAC.

State, Local, and Tribal Advisory Structures

Advisory structure innovation is not limited to the federal level. Texas HB 2060 (2023) established the state's AI Advisory Council with defined membership and reporting requirements, producing recommendations to the legislature. New York City's Algorithmic Management Policy, established under Executive Order 50 (2019) and subsequent Local Law 35 (2018, Automated Decision Systems Task Force), has evolved into a permanent review structure. California's Government Operations Agency has run GenAI advisory work under Executive Order N-12-23 (2023). Washington State's AI task force, Connecticut's AI inventory work, and Pennsylvania's AI Governing Board all offer models. At the city level, Seattle's Surveillance Advisory Working Group operates under the Surveillance Ordinance (Seattle Municipal Code 14.18) and reviews specific acquisitions before council approval. Boston, Oakland, Santa Clara County, and others have similar structures. Tribal governments including the Navajo Nation and Confederated Tribes bodies have begun establishing AI advisory work aligned to tribal sovereignty principles. Cross-state bodies, the National Governors Association, NASCIO, the National Association of State CIOs' AI working group, and the Council of State Governments AI Working Group, provide peer learning and model policy.

Common Failure Modes and How to Design Around Them

Failure Mode 1 - The advisory purgatory: a committee that meets, produces reports, and receives no action. Prevent this by requiring a documented agency response to each recommendation within a defined window (30 or 60 days) and by publishing both the recommendation and the response. Failure Mode 2 - The rubber stamp: a governance board that approves everything placed in front of it. Prevent this by requiring a written staff analysis for each decision, a public vote record in the minutes, and a routine dissent pathway. Failure Mode 3 - The captured committee: membership skewed toward regulated or vended interests. Prevent this through FACA balance requirements, transparent conflict management, and balanced public-process appointments. Failure Mode 4 - The invisible structure: work done in silent internal groups with no public record. Prevent this through published charters, meeting schedules, and annual reports. Failure Mode 5 - The performative committee: a body designed to absorb criticism rather than generate decisions. Prevent this by tying the committee to real authority (approval, veto, or at minimum mandatory response) rather than 'advice' in the weakest sense. Failure Mode 6 - The unfunded mandate: advisory bodies with no staff or budget. Prevent this by ensuring the DFO has protected time and the committee has a budget line. Failure Mode 7 - The immortal committee: a body that outlives its mission. Prevent this with sunset dates and renewal-contingent-on-demonstrated-value.

Reflection and Application

Before advancing, complete the following exercises against an agency or jurisdiction you lead or advise. (1) Draw the current AI advisory-structure map for your organization: every formal and informal body that provides AI-related input or decision. Identify gaps and redundancies. (2) For your governance board (or its equivalent), audit membership, decision rights, cadence, and public reporting against the M-24-10 expectations. Identify the gap that most undermines legitimacy. (3) For any FACA committee in your orbit, read the charter, last renewal, and last annual report. Identify a committee whose charter should be amended and draft the change. (4) Map the interface between your agency governance board and external advisory bodies (NAIAC, state AI task force, municipal oversight board). Identify the single communication channel that is most broken and fix it. (5) Design the advisory structure you would recommend for a brand-new program, say, AI in unemployment-insurance fraud detection, including decision body, external advisory committee, community consultation, and IG oversight interface. Write the charter for each.

Key Terms

FACA: Federal Advisory Committee Act (5 U.S.C. App. 2), the statute governing advisory committees that include non-federal members. DFO: Designated Federal Officer, the federal employee who manages a FACA committee's operations. NAIAC: National Artificial Intelligence Advisory Committee, chartered under the National AI Initiative Act of 2020. AI GOVERNANCE BOARD: The M-24-10-required agency decision body chaired by the CAO. CHIEF AI OFFICER COUNCIL: The interagency body composed of agency CAOs, chartered under OMB guidance. AI CoP: The Federal CIO Council's AI Community of Practice, a practitioner network. NAIIO: National AI Initiative Office, hosted at OSTP. SAOP: Senior Agency Official for Privacy, the M-24-10 privacy role typically served by the Chief Privacy Officer. GSA CoE: The General Services Administration AI Center of Excellence. SUNSHINE ACT: The Government in the Sunshine Act (5 U.S.C. § 552b), governing open meetings of federal collegial bodies. 18 U.S.C. § 208: The federal criminal conflict-of-interest statute applicable to SGEs and FACA committee members.

Related Lectures

L5 5.3.1 - AI Governance Foundations
L5 5.3.2 - AI Regulatory Design
L5 5.3.3 - Transparency: Citizens' Right to Know