• +1 (602) 922-5990
  • This email address is being protected from spambots. You need JavaScript enabled to view it.
  • Mon - Fri 8:00a - 5:00p AZ
Industry Insights banner: AI, Contractors, and Continuity of Operations, a resilience briefing for public agencies, with a domed capitol.

AI, Contractors, and Continuity of Operations: A Resilience Briefing for Public Agencies

A public agency can hold a continuity of operations plan, a set of information-technology contracts, and a growing list of artificial-intelligence use cases, and still be unprepared for the moment a contractor, a cloud service, or a model used in a public-facing decision is unavailable or untrustworthy. For state, local, tribal, and territorial governments, as well as for federal executive agencies and public pension systems, that is now a single management problem. Continuity, contractor oversight, and AI governance are often housed in different offices. The public experiences them as one service: either the agency can still perform its essential functions, fairly and safely, or it cannot.

The official federal starting point for agency use of AI has also changed, and that change should be stated accurately. On March 28, 2024, the Office of Management and Budget issued Memorandum M-24-10, Advancing Governance, Innovation, and Risk Management for Agency Use of Artificial Intelligence. On April 3, 2025, OMB issued Memorandum M-25-21, Accelerating Federal Use of AI through Innovation, Governance, and Public Trust, which in its official text "rescinds and replaces" M-24-10. M-24-10 is therefore part of the recent history of federal AI governance. It is not the current OMB memorandum. This article works from M-25-21, from FEMA's Continuity Guidance Circular as whole-community continuity guidance, and from the NIST AI Risk Management Framework as a voluntary method that agencies and their contractors can use together.

Why this matters

Essential functions in the public sector almost always run on a mix of government staff and contracted systems: benefits and licensing platforms, 911 and public-safety systems, utility billing, tax and finance systems, records, cloud hosting, and, increasingly, AI used to summarize cases, detect fraud, route service requests, or support decision-making. Public pension systems have a parallel dependency on custody, administration, investment, and cybersecurity vendors. Tribal governments may operate public-safety, health, utility, and enterprise services with a small staff and a concentrated set of contractors. In each of those settings, a contractor outage or a flawed model can interrupt a function the public cannot easily obtain elsewhere.

FEMA publishes the Continuity Guidance Circular as whole-community guidance to support the development of continuity capabilities across the nation. FEMA's continuity pages describe the Circular as a resource for organizations to integrate and coordinate continuity efforts. It is intended to be flexible and adaptable for a broad range of audiences, including but not limited to federal entities and other jurisdictions. It is guidance, not a substitute for an agency's own legal duties, and it is the official whole-community reference this article uses for continuity planning. A continuity program that never names the contractors and models that essential functions depend on is not a complete program, however carefully the plan is formatted.

This article does not restate lessons from the United Kingdom's National Resilience Report, and it does not treat that report as US regulation. The question here is narrower and domestic: how US public agencies should connect continuity of operations, contractor risk, and current federal AI governance.

What M-25-21 now requires of federal agencies

M-25-21 is dated April 3, 2025, and is directed to the heads of all Executive Branch departments and agencies, including independent regulatory agencies. It does not cover AI when it is being used as a component of a National Security System, as that term is defined in the memorandum. The memorandum directs agencies to accelerate federal use of AI by focusing on three priorities: innovation, governance, and public trust. It tells agencies to remove unnecessary barriers to responsible adoption, to develop public AI strategies, and to build governance that enables effective and safe innovation rather than acting only as a bottleneck.

Governance duties in the official text include designating Chief AI Officers, establishing an agency AI governance board, updating agency policies, and developing a generative-AI policy, each on the timelines set in the memorandum. CFO Act agencies must develop public AI strategies. Those organizational requirements matter for continuity and contracting because they create named officials and forums that can be asked a straightforward question: which AI uses support essential functions, which of those uses are high-impact, and which are delivered by contractors?

The memorandum also directs agencies to implement minimum risk-management practices for "high-impact AI," defined as AI with an output that serves as a principal basis for decisions or actions with legal, material, binding, or significant effect on: an individual or entity's civil rights, civil liberties, or privacy; access to education, housing, insurance, credit, employment, and other programs; access to critical government resources or services; human health and safety; critical infrastructure or public safety; or strategic assets or resources, including high-value property and information marked as sensitive or classified by the Federal Government. When high-impact AI is not performing at an appropriate level, agencies must have a plan to discontinue its use until compliance actions are taken. If proper risk mitigation is not possible, agencies must cease the use of the AI.

The minimum practices themselves, set out in Section 4(b) of the memorandum, include: pre-deployment testing and risk-mitigation plans, including alternative test methods when the agency does not have access to source code, models, or data; a documented AI impact assessment before deployment, updated over the lifecycle, including independent review and risk acceptance; ongoing monitoring for performance and adverse impacts; sufficient and periodic training and assessment for operators; human oversight, intervention, and accountability suitable for the use, including an appropriate fail-safe when practicable; consistent remedies or appeals for individuals affected, when appropriate; and consultation and feedback from end users and the public where appropriate. Several of those practices are difficult to perform if the AI is delivered as a closed contractor service and the contract does not provide testing access, monitoring information, or a clear discontinuation path.

M-25-21 is an OMB memorandum to federal executive agencies. It does not, by its own terms, bind a city, a county, a tribal government, or a public pension system. Those organizations can still use its definition of high-impact AI and its minimum practices as a structured way to ask better questions of their own uses and vendors. They should not describe M-25-21 as their legal requirement unless a separate authority makes it so.

Continuity, contractors, and the same essential functions

A continuity of operations plan should identify essential functions, orders of succession, delegations of authority, essential records, alternate facilities or modes of operation, communications, and tests and training. In a modern agency, several of those elements are performed by contractors or hosted in contractor environments. If the benefits system, the records system, or the public-safety platform is unavailable, the essential function is unavailable, even if the agency's own building and staff are intact.

That is why contractor inventory and continuity planning should be one picture. The agency should know which contracts support essential functions, what service and recovery commitments those contracts contain, how incidents are notified, whether there is an alternate path, and how the agency would operate if the contractor cannot perform. Exercises should include contractor and cloud failure, not only facilities and weather. Procurement and contract administration should be able to produce, for those essential contracts, the rights needed to test, to obtain information, and to transition. For a broader discussion of proving third-party resilience, concentration in shared platforms, and what to do when a critical provider cannot recover, see Beyond Due Diligence: Proving Third-Party Resilience, The Hidden Risk of Concentration, and When a Critical Provider Cannot Recover.

Public pension systems and tribal governments often have fewer staff and a shorter list of critical vendors. Proportionate does not mean undocumented. A short inventory of essential functions, material contracts, and material AI uses, with a recovery path and a named owner, is a better governing-body report than a long plan that never mentions the contractor.

AI governance when the model is contracted

Many agency AI uses are not built by government data scientists. They are features inside a licensed platform, a cloud service, or a professional-services engagement. M-25-21 anticipates that problem in its testing practice: if an agency does not have access to underlying source code, models, or data, it must use alternative test methods, such as querying the service and observing outputs or providing evaluation data to the vendor and obtaining results. Contracts for high-impact uses should make those methods possible, and should address monitoring, human oversight, notice to affected people where appropriate, and a path to discontinue the use.

The National Institute of Standards and Technology published the Artificial Intelligence Risk Management Framework (AI RMF 1.0), NIST AI 100-1, on January 26, 2023. It is voluntary, rights-preserving, and organized around Govern, Map, Measure, and Manage. It is not sector-specific, which makes it usable by federal agencies, by SLTT and tribal governments, by public pensions, and by the contractors that serve them. It describes trustworthy AI as valid and reliable, safe, secure and resilient, accountable and transparent, explainable and interpretable, privacy-enhanced, and fair, with harmful bias managed. Those characteristics line up with the public-trust priority in M-25-21 and with the fairness and safety concerns in the high-impact definition.

ISO/IEC 42001:2023, published in December 2023, provides a management-system standard for organizations that provide or use AI. Certification is voluntary and is not performed by ISO. Agencies should not require a contractor certification they cannot evaluate, but they can ask contractors how policies, roles, risk assessment, and lifecycle control are organized, and they can use the same structure internally.

See also Connecting AI Governance and Resilience and From AI Principles to Accountability.

Where the three issues meet

The meeting point is an essential function that cannot pause without harming the public: a benefits determination, a public-safety workflow, a utility or tax service, a pension payment, or access to a critical government resource. If the contractor is down, or if a high-impact model is no longer performing, the agency still has to perform the function or discontinue the AI and use a workable alternative. Continuity plans, contract files, and AI inventories that cannot be read together will not support that decision.

What organizations should do

Name essential functions first, then the contractors and AI uses that support them. Use FEMA's Continuity Guidance Circular as a whole-community planning resource, and put contractor and model dependencies into the plan, the training, and the exercise cycle.

If you are a federal executive agency, work from M-25-21 as the current OMB memorandum. Do not cite M-24-10 as if it were still in force. Identify high-impact AI uses, including contracted uses, and implement the memorandum's minimum practices, including a path to discontinue a use that is not performing.

If you are a state, local, tribal, or territorial agency or a public pension system, do not claim M-25-21 as your binding rule unless it is made applicable by another authority. Do use its high-impact definition and minimum practices, and the NIST AI RMF, as a structured way to govern your own uses and vendors.

Write contracts for essential and high-impact services so that testing, monitoring, incident notification, human oversight, and transition or discontinuation are possible. Exercise the loss of a contractor or a model, and record findings with owners and dates.

Give the governing body or senior officials one report: essential functions, material contracts, material AI uses, exercise results, and open issues. See also From Plans to Proven Readiness and Operational Resilience: Where to Start.

How OpResONE can help

OpResONE helps public agencies and public pension systems strengthen continuity of operations, contractor oversight, and AI governance as a single management problem, including dependency mapping and the design of exercises and official reporting. Our work is independent. We have no affiliation with OMB, FEMA, NIST, or ISO, and this article is not an official interpretation of their memoranda, guidance, or standards.

We bring BRILLIANCE to RESILIENCE®.

Related resources

Request a Strategy Call

If you want a focused reading of how contractor oversight and AI governance line up with the continuity of your essential functions, talk with OpResONE about where to start.

Request a Strategy Call

Sources