Many organizations now run two important programs side by side without connecting them. One is AI governance: the policies, committees, and risk assessments that decide how artificial intelligence may be developed, bought, and used. The other is operational resilience: the discipline of keeping critical services running through disruption. They are often staffed by different people, report to different executives, and use different vocabularies.
That separation made sense when AI sat at the edges of the business, in pilots and productivity tools. It makes less sense now that AI models help route customer calls, flag fraud, schedule maintenance, draft communications, and prioritize work queues. When AI becomes part of how a critical service is delivered, its failure becomes a resilience problem, and resilience planning that ignores it has a blind spot.
This article explains why the two disciplines need each other and how leaders can connect them without building another layer of bureaucracy.
"There will be a human in the loop." It is one of the most common reassurances offered when organizations deploy artificial intelligence, and one of the least examined. A person somewhere in the process is expected to catch errors, prevent harm, and take responsibility. Yet in practice, that person may be reviewing hundreds of AI recommendations an hour, may lack the information to judge them, may have no real authority to overrule them, and may have learned through experience that the AI is usually right.
In those conditions, human oversight becomes a formality. The human signs off, the AI effectively decides, and accountability becomes harder to locate when something goes wrong.
Meaningful oversight is achievable, but it has to be designed. This article explains how to define review and intervention points that work: where to place them, what the people in them need, and how to know whether oversight is real.
Over the past several years, many organizations have published statements of AI principles. They commit to fairness, transparency, privacy, safety, and human oversight. These statements are a reasonable starting point. They signal intent to employees, customers, and regulators, and they give teams a shared vocabulary.
But principles do not govern anything on their own. A commitment to fairness does not tell a product team which tests to run before launching a credit model. A commitment to transparency does not say who decides what customers are told when AI is involved in a decision. A commitment to human oversight does not specify who can stop a system that is behaving badly. Between the principle and the practice sits a gap, and that gap is filled either by deliberate accountability or by assumptions.
This article explains how to turn AI principles into practical accountability that works across the full lifecycle of AI systems, from the first idea to retirement.
A life-sciences organization can have validated equipment, approved procedures, and a quality unit and still be unable to release a batch, complete a critical test, or keep a multi-site laboratory network running if a contract manufacturer, a specialized testing lab, a cloud laboratory system, or a model used in development or quality cannot perform. In that moment the problem is not only a supplier-management finding. It is a continuity problem, a quality problem, and, where models are in the path, an AI-governance problem.
Life sciences is not a single regulatory silo. Drug and biological product manufacturers, blood and biologics laboratories, and medical-device firms live under different Food and Drug Administration frameworks, and many also sell or operate in the European Union. This article does not treat those frameworks as interchangeable, and it does not invent client or facility examples. It asks how third-party risk and AI should be governed so that testing, manufacturing, and release can continue when a partner or a model fails.
Organizations want their customers, employees, regulators, and boards to trust the AI systems they use. Trust, however, cannot be requested. It has to be earned, and in a governance context it is earned through evidence: documented, verifiable information showing that an AI system was designed responsibly, tested appropriately, operates as intended, and is overseen effectively.
Many organizations have pieces of that evidence. A data science team has validation results. A vendor has supplied a technical summary. A risk team has completed an impact assessment. A monitoring dashboard tracks some performance metrics. But the pieces often sit in different places, follow different standards, and stop at the moment of deployment. When someone asks a direct question, such as "How do we know this model is still performing fairly?", assembling an answer takes weeks, if it can be done at all.
This article sets out how to connect testing, monitoring, and assurance into a continuous chain of evidence that supports trust in AI over its full life.
A technology firm can hold a current service-organization report, a customer contract with an uptime target, and a product roadmap for artificial intelligence, and still be unprepared for the afternoon it becomes someone else's operational incident. For many software, cloud, and SaaS companies, that is the real resilience question. The firm is not only a user of vendors. It is a third party to banks, health systems, utilities, agencies, and other businesses that have already defined the firm's platform as a critical service.
That inversion changes the work. Operational resilience is not only the firm's ability to keep its own operations running. It is the firm's ability to keep the service its customers treat as essential available, recoverable, and understandable when a subprocessor, a region, or a model fails. Third-party risk runs in both directions: the firm must manage its own suppliers, and it must be able to show customers, and in some cases supervisors, that it can be managed as a supplier. AI governance is part of the same promise whenever a model sits in the product or in the pipeline that keeps the product running.
Banks and credit unions can keep a complete vendor file and still be unprepared for the afternoon a core processor, a cloud platform, or a model they do not fully control stops supporting a critical service. The useful question for a board is not whether due diligence checklists exist. It is whether the institution can keep important financial services available when those dependencies fail, and whether directors can see that picture in time to challenge it.
Two official developments make that question sharper. In June 2023, the Office of the Comptroller of the Currency, the Board of Governors of the Federal Reserve System, and the Federal Deposit Insurance Corporation issued the Interagency Guidance on Third-Party Relationships: Risk Management, which remains the current interagency view for the banking organizations those agencies supervise. On September 15, 2026, the same agencies, now joined by the National Credit Union Administration, published proposed replacement guidance and asked for comment by November 16, 2026. Until any final guidance is issued, the 2023 interagency guidance, and for credit unions the NCUA's existing third-party letter, are the documents institutions should work from.
At the same time, institutions are putting artificial intelligence into credit, fraud, operations, and customer service, often through vendors. A model that cannot be explained, tested, or replaced is not only an AI-governance issue. It is a third-party issue and an operational-resilience issue. This article looks at those three questions together, especially for community banks and credit unions.
ISO/IEC 42001:2023, Information technology — Artificial intelligence — Management system, is the international standard for an artificial intelligence management system (AIMS). ISO published the first edition in December 2023. It specifies requirements, and provides guidance, for establishing, implementing, maintaining and continually improving an AIMS, and is intended for any organization that provides or uses products or services that utilize AI systems.
This article explains what ISO 42001 certification is, who typically needs it, how individual competence — including implementer and auditor roles — differs from organizational certification, and how to prepare. It draws on ISO's published pages and on the published structure of the standard. OpResONE has no affiliation with ISO. This article is not an official interpretation of the standard, and it is not a substitute for a licensed copy.
A health system's continuity plan can name generators, evacuation routes, and clinical documentation and still leave a quieter gap: the electronic health record, the imaging archive, the staffing platform, or a clinical decision tool that is owned, hosted, or updated by someone else. When one of those services fails, the event is not only an information-technology outage. It can become a care-continuity, privacy, and AI-governance event at the same time.
Healthcare already has official requirements that touch this gap, even if they were not written as an "AI and vendor resilience" program. The Centers for Medicare & Medicaid Services emergency preparedness condition of participation for hospitals, at 42 CFR 482.15, requires an all-hazards program, a plan reviewed at least every two years, and exercises at least twice a year. The HIPAA Security Rule requires covered entities and their business associates to safeguard the confidentiality, integrity, and availability of electronic protected health information, including through a contingency plan. Neither rule is a complete method for governing clinical AI or a complex vendor estate. Both are reasons not to treat those issues as optional.
This article is about that combined agenda: how hospitals and health systems govern contracted services and AI tools so that care can continue when a vendor or a model fails. It is a reading of the official requirements as they now stand, and of the questions leaders should ask of the third parties and models that sit in the path of care.
Many organizations can produce a list of courses people have attended. Fewer can show that the people who run business continuity, operational resilience or AI governance can actually do the work those systems require. Attendance is not competence. Management system standards treat that distinction as a requirement, not as a training preference.
This article explains why competency development matters, what ISO 22301:2019 and ISO/IEC 42001:2023 require in clause 7.2, how a simple competency framework helps, and how to move beyond attendance certificates to demonstrated competence. It paraphrases the published requirements; audit interpretations should be made against a licensed copy of each standard. OpResONE has no affiliation with ISO, and this article is not an official interpretation of the standards.
Energy and water utilities are often asked to show a continuity plan, a cyber program, and a vendor list. The harder question is whether a utility can keep essential service flowing when a supplier, a control-system vendor, or a model used in operations fails. For many organizations those three issues sit under different managers and different regulatory labels, even though they describe the same operating problem.
The official landscape is split, and it is a mistake to blur the split. Applicable users, owners, and operators of the Bulk Electric System are subject to mandatory NERC Critical Infrastructure Protection reliability standards, which the Federal Energy Regulatory Commission approves and which include supply-chain and recovery obligations. Community drinking-water systems that serve more than 3,300 people are subject to America's Water Infrastructure Act section 2013, which amended section 1433 of the Safe Drinking Water Act and requires a risk and resilience assessment and an emergency response plan, recertified on a five-year cycle. Wastewater systems are encouraged to plan; they are not required to certify those products to EPA. Public power may fall under CIP, AWIA, both, or neither, depending on registration and on whether the organization also operates a community water system.
This article treats that split as the starting point, then asks how third-party risk and AI used in operations should be governed so that recovery, supply-chain control, and resilience assessments describe the same set of critical dependencies.
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.