Governing Body and Executive Oversight of Healthcare AI in Viet Nam: What Healthcare Leaders Should Review, Escalate and Require Action On
IMPLEMENTATION INSIGHTSGOVERNANCE, LEADERSHIP & ACCOUNTABILITY
Governing Body and Executive Oversight of Healthcare AI in Viet Nam: What Healthcare Leaders Should Review, Escalate and Require Action On
IMPLEMENTATION INSIGHT
Last reviewed: 7 October 2026
As healthcare organizations adopt more AI, responsibility cannot remain entirely with IT teams, individual clinical departments or an AI governance committee. Operational teams may evaluate evidence, perform local validation, review vendors, monitor performance and investigate incidents, but some questions are fundamentally organizational. Which AI uses could materially affect patient safety, quality or institutional trust? Is AI actually improving care or operations? Are recurring safety concerns being resolved? Is governance capacity keeping pace with adoption? Which risks exceed delegated authority? When should senior leadership intervene?
These questions belong at a different level of governance.
Joint Commission's current voluntary Responsible Use of AI in Healthcare (RUAIH) certification reflects this shift toward organization-level responsibility. RUAIH assesses whether healthcare organizations have formal governance, organization-wide policy, appropriate resources, data safeguards, risk and bias controls, lifecycle monitoring, safety-event reporting, transparency and education in place. It focuses on responsible organizational use of AI rather than certification of individual AI products [6].
Earlier Joint Commission–CHAI guidance made the leadership connection particularly explicit. It recommended a formal AI governance structure, a mechanism for keeping the healthcare organization's governing body informed about AI uses, outcomes and potential adverse events, and regular updates to the organization's fiduciary board on AI use and outcomes [8].
For healthcare organizations in Viet Nam, these international concepts need to be adapted carefully. Public hospitals, private hospitals, university hospitals, corporate healthcare groups, specialty clinics and other healthcare organizations can have very different ownership, governance and management structures. “Governing body” in this article is therefore used as a functional governance concept rather than as a universal statutory title under Vietnamese law. Depending on the organization, that function may sit with a board, owner or corporate governance body, hospital director and senior leadership, or another body with ultimate organizational oversight.
The practical question is not whether every healthcare organization in Viet Nam should create an “AI Board Committee.” It is what information about AI needs to reach senior organizational leadership, what decisions should remain delegated, and what situations require leadership intervention.
Governing-body oversight is not the same as operational AI governance
One of the first mistakes organizations can make is sending every AI decision upward. A governing body or senior executive team should not normally decide whether a particular model threshold should be adjusted, whether one validation metric is acceptable or whether a routine software update passes technical testing. Those decisions belong to people with the appropriate clinical, technical, quality, legal, regulatory and data expertise.
Leadership oversight serves a different purpose. Operational AI governance asks whether a particular system is appropriate, safe and properly controlled. Executive management oversight asks whether management structures, resources and actions are sufficient to keep AI use safe and effective across the organization. Governing-body oversight asks whether the organization is using AI consistently with its responsibilities, strategy, organizational risk boundaries, patient-safety obligations and accountability to the people it serves.
A useful organizational model is therefore:
operational AI governance → executive management oversight → governing-body assurance and accountability
This is a practical organizational model proposed here, not a statutory hierarchy under Vietnamese law. Its value lies in distinguishing levels of decision-making. When every AI decision goes to the top, governance becomes slow and superficial. When nothing material reaches the top, organizational accountability becomes weak.
Viet Nam already places healthcare quality and patient safety within leadership responsibility
Healthcare AI leadership should not be treated as an entirely new concept disconnected from existing Vietnamese healthcare governance.
Circular No. 19/2013/TT-BYT remains recorded as in force. It establishes patient-centeredness, continuous quality improvement and evidence-informed quality decision-making as principles of hospital quality management. It also places responsibility for hospital quality with the hospital director and establishes a quality-management structure that includes a Hospital Quality Management Council chaired by the director, a quality-management unit and a quality network [5].
The current consolidated Law on Medical Examination and Treatment similarly provides an important patient-safety foundation. Article 71 states that prevention of medical incidents is based on identification, reporting, causal analysis and preventive action to avoid recurrence and improve healthcare quality and patient safety. It also places responsibility for medical-incident prevention on the head of the healthcare establishment and people working within it [4].
These provisions do not create an AI-specific governing-body oversight model. They do, however, reinforce an important principle: where AI materially affects healthcare quality, clinical work or patient safety, it should not be treated as an issue belonging only to technology teams.
AI governance should connect with the organization's existing leadership responsibilities for quality and safety.
Vietnamese AI law creates organizational responsibilities, but not a universal AI-board model
Law No. 134/2025/QH15 on Artificial Intelligence has been effective since 1 March 2026, and Decree No. 142/2026/NĐ-CP has been effective since 1 May 2026 [1,2]. The framework establishes responsibilities for AI providers and deployers according to system risk and use, including lifecycle obligations for high-risk AI.
Decision No. 33/2026/QĐ-TTg, effective since 15 August 2026, separately establishes Viet Nam's current List of High-Risk Artificial Intelligence Systems [3].
For healthcare organizations acting as deployers, relevant responsibilities may include operating systems within their intended conditions, organizing risk management during deployment and use, monitoring for deviations, risks or incidents, maintaining appropriate human supervision and intervention, applying controls within their authority, and coordinating with providers when reclassification, significant changes or incidents require joint action [1,2].
The framework also establishes formal reporting and handling responsibilities for serious AI incidents. These duties matter to leadership because a serious AI-related event can create clinical, regulatory, operational, financial and reputational consequences requiring coordinated organizational action.
Current Vietnamese AI law should nevertheless not be described as requiring all healthcare organizations to create a board-level AI committee or prescribing the leadership model proposed in this article. Operational legal responsibility and organizational leadership oversight are related, but they are not the same thing.
Leadership should see organizational AI exposure, not only individual projects
Executives cannot oversee what they cannot see.
A useful leadership view begins with basic organizational visibility: how many AI-enabled capabilities are deployed, being piloted, under validation, awaiting approval, restricted, suspended or being retired? Which departments and sites use them? Which are patient-facing? Which materially influence clinical decisions or prioritization? Which process sensitive information? Which can initiate actions rather than simply provide information?
Leadership does not need a technical inventory containing every model parameter. It needs a portfolio-level view of organizational exposure.
A healthcare organization using two controlled AI tools has a different governance problem from one using dozens of AI-enabled capabilities across radiology, pathology, endoscopy, clinical documentation, patient communication, quality management and administration.
Joint Commission's current RUAIH health-system pathway illustrates this organization-level perspective. Where AI activities are centrally governed across participating healthcare organizations, the system pathway expects common governance, processes and policies for AI use [6]. This is a U.S. voluntary certification model rather than a Vietnamese requirement, but it illustrates an important principle: as AI scales, leadership eventually needs to oversee the organization or system around the technologies, not just individual projects.
The detailed methods for segmenting, prioritizing and managing an AI portfolio belong to portfolio governance. At governing-body and executive level, the priority is visibility into where material organizational exposure is accumulating.
Leaders should know which AI uses matter most
A long inventory is not useful if leadership cannot distinguish material exposure from routine low-consequence use.
Executives and governing bodies should therefore receive a risk-proportionate view of AI. Particular attention may be warranted where AI is close to consequential clinical decisions, communicates directly with patients, has substantial action authority, operates at large scale, affects vulnerable populations, processes sensitive information, creates difficult-to-reverse consequences or depends on important external infrastructure.
Novelty and uncertainty can also matter. A system with limited real-world evidence may warrant leadership visibility even before a serious safety signal emerges.
This does not mean that a governing body should independently perform statutory AI risk classification. Classification follows the applicable Vietnamese legal framework and the responsibilities of providers and deployers. Leadership oversight asks a different question:
Which AI uses could materially affect the organization if governance fails?
That keeps senior oversight focused on organizational consequence rather than regulatory or technical micromanagement.
Leadership should review whether AI is delivering value, not merely whether it has been deployed
Technology adoption is easy to measure. Value is harder.
An executive dashboard that reports only the number of AI systems deployed, users trained or departments using AI can create a misleading picture of progress. Leadership should ask whether important AI uses are actually achieving the healthcare or operational outcomes for which they were introduced.
For a deterioration model, the relevant question may be whether timely clinical review or meaningful patient outcomes improve. For ambient documentation, leaders may want to know whether documentation burden decreases without unacceptable accuracy problems. For imaging prioritization, the question may be whether urgent cases receive faster review without creating unacceptable delays elsewhere.
Administrative AI may produce efficiency, lower cost or reduced staff burden, but these benefits should still be considered alongside errors, downstream workload and unintended consequences.
Leadership should therefore move from asking “Are we using AI?” to asking “Is this AI producing enough benefit to justify the risks, costs and governance resources associated with it?”
Safety signals should reach leadership before they become crises
Executives should not first learn about an AI safety problem when it becomes a serious incident.
Operational teams need enough authority to manage routine errors and near misses without escalating every case. Leadership, however, should receive information about patterns and material signals, such as recurring near misses, significant performance deterioration, repeated overrides, patient complaints, persistent subgroup differences, major workflow problems or unresolved vendor issues.
Joint Commission–CHAI guidance specifically called for keeping the governing body informed about AI uses, outcomes and potential adverse events [8]. Current RUAIH similarly includes lifecycle monitoring and internal and external reporting processes for health-AI safety-related events [6].
Leadership oversight should therefore receive signal summaries rather than raw incident counts. One near miss may require operational review. A recurring pattern across several departments or sites may indicate an organizational problem.
Leaders need clear escalation criteria
Good governance depends on knowing which issues remain delegated and which move upward.
Executive or governing-body escalation may be appropriate when an AI issue involves actual or potential serious patient harm; material regulatory exposure; inability to maintain meaningful human oversight; persistent performance deterioration despite corrective action; significant cybersecurity or privacy risk; major changes in intended use or action authority; organization-wide or multi-site consequences; material vendor or infrastructure dependency; unresolved disagreement among governance functions; or residual risk beyond delegated management authority.
These are organizational examples, not universal legal thresholds.
Each healthcare organization should define its own leadership escalation criteria according to applicable law, organizational structure, service profile, governance arrangements and the level of risk management is authorized to accept.
The important principle is that escalation should be designed before the crisis rather than negotiated during it.
Material residual risk should become visible at the appropriate leadership level
No healthcare AI system will be risk-free.
After evidence review, local validation, workflow controls, training and monitoring are in place, some uncertainty or residual risk will remain. Operational governance may be authorized to manage or accept defined lower-level residual risks within delegated limits, while more consequential unresolved risk should become visible at an appropriate leadership level.
Leadership should understand what material residual risk remains, why it has not been eliminated, which controls mitigate it, who owns the response, how long the organization intends to accept the current position, how the risk will be monitored and what would trigger reconsideration.
The purpose is not to make executives sign every risk register. It is to prevent material organizational risk from being silently accepted at too low a level.
A useful principle is that authority to manage or accept defined residual risk may be delegated within appropriate limits, but accountability for a functioning governance system still needs to remain clear.
Governing bodies should oversee exception patterns without becoming the exception committee
Organization-wide AI policies will inevitably have exceptions. A controlled research project, temporary workaround or unusual operational need may justify deviation from an internal rule.
Most exceptions should not require governing-body approval.
Leadership should instead receive visibility into material patterns of exceptions. If several departments repeatedly request the same exception, the underlying policy may be unrealistic. If a temporary exception continues for months, it may no longer be temporary. If a high-consequence AI system repeatedly operates under temporary exemptions, leadership may need to require a more durable resolution.
Useful executive questions include whether exceptions are increasing, whether high-consequence exceptions recur, whether temporary exceptions expire as intended, whether compensating controls are effective and whether exception patterns reveal a deeper governance or resource problem.
This turns exceptions into organizational learning rather than a quiet route around policy.
Significant changes deserve leadership visibility when they alter organizational exposure
AI systems evolve. A vendor may introduce a new model, expand action authority, change data processing, add patient-facing functionality or alter important infrastructure dependencies. A healthcare organization may expand a system to new populations, specialties or sites.
Not every change needs executive review, but leadership should be informed when a change materially alters organizational exposure.
For high-risk AI, Viet Nam's framework contains specific requirements relating to significant changes, conformity reassessment and reclassification depending on the circumstances and respective roles of providers and deployers [1,2]. Organizational change-control processes should therefore be capable of distinguishing routine technical updates from material changes that alter prior risk, validation or approval assumptions.
A useful leadership question is:
“Has this system become meaningfully different from the system we previously approved?”
Major vendor and infrastructure dependencies belong on the leadership agenda
Healthcare AI can create organizational dependencies that are easy to miss when products are reviewed one at a time.
Several clinically important AI systems may depend on the same cloud provider, foundation model or external API. A vendor may become embedded across several important workflows. A single external outage or commercial change could therefore affect multiple services.
Leadership does not need to manage every third-party relationship directly, but it should have visibility where concentration or dependency becomes materially important. Relevant questions include how much clinically important AI depends on the same external service, whether alternatives or fallback processes exist, whether important functionality can change outside the organization's direct control, whether data and records can be recovered on exit, and what would happen if a critical provider stopped supporting the service.
CHAI's 2026 Governance Playbooks include Third Party Management as a core governance domain and are designed so that responsible-AI controls can be integrated into existing organizational processes [7].
The detailed analysis of concentration risk belongs to portfolio and third-party governance. At leadership level, the question is whether a dependency has become significant enough to require organizational action.
Leadership should review aggregate workload and human-factor risk
AI can appear safe one system at a time and still create problems in combination.
Five individually acceptable AI systems may each generate a manageable number of notifications for nursing staff, while the combined alert burden becomes unacceptable. Several generative systems may each require modest human verification but collectively create substantial hidden workload.
Executives should therefore ask not only “Is each AI safe?” but also “What is the combined effect of AI use on people and workflows?”
Relevant signals may include alert burden, correction work, additional handoffs, documentation workload, staff complaints, recurrent workarounds or increasing operational dependence on AI.
This is one reason AI leadership oversight should remain connected with quality management, workforce planning and patient safety rather than sitting exclusively within digital strategy.
Bias and equity concerns need leadership attention when they become organizational issues
Operational teams should evaluate relevant subgroup performance and investigate potential bias. Leadership does not need to review every statistical subgroup analysis.
Persistent disparities should nevertheless be escalated when they materially affect patient access, quality, safety or organizational commitments.
CHAI's Governance Playbooks include risk and impact assessment as a core domain, while RUAIH assesses organizational processes for identifying, addressing and mitigating AI risk and bias [6,7].
Leadership should therefore know whether important AI uses perform acceptably for relevant populations, whether inequities have appeared in real-world outcomes or access, whether corrective actions are effective and whether use should be narrowed where acceptable performance cannot be demonstrated.
The aim is not to make the governing body a statistical review committee. It is to ensure that unresolved inequity does not remain buried in technical reports.
Data, privacy and cybersecurity should be reported as organizational exposure
A governing body does not need to review encryption settings or database architecture.
It does need to know when AI creates material exposure involving sensitive patient data, important cross-border arrangements, significant cybersecurity vulnerabilities, unexpected secondary data use or major third-party dependencies.
Leadership reporting should therefore summarize the critical data involved, where important data flows leave direct organizational control, whether significant privacy or cybersecurity concerns remain unresolved, whether important vendor obligations are unclear and whether adequate incident-response and fallback arrangements exist.
Senior oversight is similar to other areas of organizational governance: leadership does not perform every technical control, but it needs assurance that material exposure is understood and managed.
Governing bodies should ask whether human oversight is actually credible
For AI uses that depend on clinical human oversight, an organizational policy may state that appropriately qualified professionals retain relevant decision authority. Leadership should ask whether real working conditions make that oversight meaningful.
Do users understand the system's important limitations? Can they challenge or override outputs when appropriate? Does workload allow adequate review? Do interfaces communicate uncertainty sufficiently? Is safe fallback available when AI is unavailable or unreliable? Are staff reporting automation bias, alert fatigue, overreliance or difficulty reviewing outputs?
Human oversight is not merely a feature designed into software. It is an organizational capability that depends on staffing, competency, workflow, interface design and culture.
Where meaningful oversight cannot be maintained, leadership may need to require redesign, restrict the use case or suspend use.
Leadership should know whether governance has enough resources
AI governance requires people, expertise and time.
Someone must conduct intake, review evidence, perform validation, monitor systems, investigate incidents, manage vendor issues, reassess changes, train users and document decisions.
A governance program can look sophisticated on paper while depending on a small number of people carrying these responsibilities informally on top of their normal work.
CHAI's Governance Playbooks explicitly include Organizational Resources as a core domain alongside AI Policy and Organizational Structures [7].
Leadership should therefore understand whether governance capacity is keeping pace with AI adoption. Useful indicators may include the number of systems being governed, review backlog and review time, missing expertise, completion of monitoring obligations and accumulation of unresolved actions.
A shortage of governance capacity is itself an organizational risk.
Executive enthusiasm should not weaken credible challenge
AI initiatives may carry strategic prestige. Senior leaders may sponsor flagship deployments, announce partnerships or commit significant resources.
That creates a predictable governance tension: people responsible for reviewing safety and evidence still need to be able to challenge a project when senior leaders strongly support it.
Leadership should therefore distinguish sponsorship from assurance.
The executive sponsoring an AI initiative may be responsible for achieving strategic value, while quality, patient safety, legal, privacy and governance functions need enough independence and authority to identify reasons the organization should slow, modify, restrict or stop the project.
A mature leadership culture welcomes credible challenge.
A useful organizational principle is:
“We want this AI to succeed if the evidence shows that it improves care or operations without creating unacceptable risk.”
Leadership should know which AI systems have been restricted, suspended or retired
Organizations often celebrate adoption and pay far less attention to withdrawal. That creates distorted governance.
Executives and governing bodies should receive visibility into significant decisions to restrict, suspend or decommission AI, particularly where those decisions reveal broader organizational lessons.
Leadership should understand why a significant system was stopped or narrowed: inadequate performance, workflow problems, failure to demonstrate expected benefit, an unacceptable change in vendor or system behavior, unresolved privacy or cybersecurity concerns, or patient-safety issues.
Negative decisions can be valuable organizational evidence. Stopping ineffective or unsafe AI is not necessarily a sign of governance failure.
Sometimes it is evidence that governance worked.
The governing body should receive assurance, not raw data
One of the practical challenges in leadership oversight is deciding what should appear on an executive or governing-body dashboard. Hundreds of technical metrics will usually obscure rather than improve oversight.
A concise Healthcare AI Leadership Oversight Dashboard can instead organize information around eight areas. Portfolio visibility shows deployed, piloted, restricted, suspended and retired systems, with emphasis on high-consequence uses. Safety signals summarize material incidents, serious near misses, recurring patterns and unresolved corrective actions. Performance and outcomes show whether major AI uses continue to achieve their intended clinical or operational results. Material and residual risk requiring leadership attention identifies significant unresolved risk, important exceptions and decisions beyond delegated authority.
Significant changes highlight major changes in functionality, action authority, use case, data or deployment context. Data, privacy, cybersecurity and third-party exposure summarize unresolved material issues and important dependencies. Governance capability shows staffing, expertise, review backlog, monitoring completion and resource gaps. Actions and decisions identify systems requiring remediation, additional investment, restriction, escalation, suspension or strategic reconsideration.
The purpose is not to create another certification checklist. It is to provide leadership with enough information to make organizational decisions.
This Healthcare AI Leadership Oversight Dashboard is proposed here as a practical framework. It is not a requirement under Vietnamese law, a Joint Commission standard or a CHAI template.
Escalation should lead to a decision or required action
Leadership oversight becomes performative if material problems are presented repeatedly without a decision.
When an issue reaches executive or governing-body level, leadership should determine what action is required and who owns it. Possible actions may include requesting additional validation, strengthening monitoring, modifying workflow, increasing resources, renegotiating vendor arrangements, restricting a use case, resolving a privacy or cybersecurity issue, requiring corrective action, suspending use or asking management to return with further evidence.
Significant actions should have an accountable owner, appropriate timeframe and follow-up mechanism.
The principle is simple:
oversight is not complete when leadership has been informed; it is complete when the organization has responded appropriately and verified that response.
Governing-body reporting should be risk-proportionate
A small specialty clinic using two limited AI tools does not need the same reporting infrastructure as a large multi-hospital health system operating dozens of clinical and administrative AI capabilities.
A smaller organization may appropriately integrate AI oversight into existing executive, quality and patient-safety meetings. A large healthcare group may need structured executive oversight, periodic governing-body reporting and centralized portfolio visibility.
Joint Commission's current RUAIH framework similarly distinguishes single healthcare organizations from healthcare systems. Where a system applies through the health-system pathway and AI is centrally governed, participating organizations are expected to operate under common AI governance, processes and policies [6].
The objective is not to maximize governance bureaucracy. It is to ensure that material organizational AI exposure reliably reaches the level of leadership capable of acting on it.
A practical Governing Body and Executive AI Oversight Framework
Healthcare organizations in Viet Nam can structure senior AI oversight around a relatively simple sequence:
set direction → delegate authority → maintain visibility → review material risk and outcomes → receive assurance → escalate exceptions and weak controls → require action → verify completion → learn and adjust
Set direction means establishing organizational principles, policy and boundaries for AI use. Delegate authority means ensuring operational decisions are made at the appropriate level rather than unnecessarily pushed upward. Maintain visibility means knowing where important AI is being used and where material exposure is increasing.
Review material risk and outcomes means looking beyond deployment activity to quality, patient safety, value and unintended consequences. Receive assurance means asking whether governance controls work in practice rather than merely exist on paper. Escalate exceptions and weak controls means identifying issues that exceed delegated authority or remain unresolved.
Require action means turning leadership oversight into accountable decisions. Verify completion means confirming that required actions were implemented and effective. Learn and adjust means using incidents, outcomes and governance experience to improve policy, resources, delegation and future decisions.
This Governing Body and Executive AI Oversight Framework is proposed as a practical organizational model. It is not an official Vietnamese governance framework or a requirement to create a separate AI committee at governing-body level.
What healthcare leaders should avoid
Leadership oversight can fail in predictable ways. Technology theatre occurs when executives see impressive demonstrations but receive little information about safety, outcomes or unresolved risk. Metric overload occurs when technical data become so extensive that important signals disappear. Approval centralization pushes operational decisions upward and creates bottlenecks without improving accountability.
Another failure is governance delegation without assurance, where leadership assumes that the existence of an AI committee proves the organization is well governed. Success-only reporting highlights deployment and benefits while failed pilots, near misses, suspensions and negative evaluations remain invisible. Finally, innovation bias occurs when credible concerns raised by quality, safety or frontline teams are interpreted as resistance because leadership is already committed to the project.
Good oversight requires a different balance: strategic support for innovation combined with credible challenge and evidence-based accountability.
What this means for healthcare organizations in Viet Nam
As healthcare AI adoption expands, senior leadership will increasingly need to govern something more complex than individual technologies. They will need to govern an organizational capability.
Executives and governing bodies do not need to become AI engineers. They need to understand where AI materially affects healthcare, whether governance systems are functioning, whether benefits justify risks and resources, whether important safety concerns are being resolved, and whether management acts appropriately when evidence changes.
For hospitals in Viet Nam, this fits naturally with an existing principle of healthcare quality governance: quality and patient safety ultimately require leadership responsibility rather than delegation to a single department. Circular No. 19/2013/TT-BYT places responsibility for hospital quality with the hospital director, while the current Law on Medical Examination and Treatment places responsibility for medical-incident prevention on the head of the healthcare establishment together with people working within it [4,5].
AI does not remove those leadership responsibilities. It adds a new source of organizational risk and opportunity that leadership must be able to see.
The practical transition is therefore:
from approving AI projects → to governing organizational AI exposure → to receiving assurance → to requiring action when risk, evidence or outcomes demand it
The strongest leadership model is not one in which executives personally decide every AI question. It is one in which the right decisions are made at the right level, material issues reliably reach leadership, accountability remains clear, and the organization can act before weak governance becomes patient harm.
That is what governing-body and executive oversight of healthcare AI should ultimately achieve.
Professional education notice
This article is intended for professional education and knowledge exchange. It does not constitute individualized legal, corporate-governance, regulatory, clinical, employment, data-protection, medical-device, accreditation or quality-management advice.
The term governing body is used functionally in this article to describe the body or organizational level with ultimate governance oversight. Healthcare organizations in Viet Nam have different legal forms, ownership structures and management arrangements; the appropriate governing body, leadership structure and delegation of authority should therefore be determined according to the specific organization and applicable law.
Nothing in this article should be interpreted as creating a universal requirement under Vietnamese law for healthcare organizations to establish a board-level AI committee or for a governing body to approve individual AI systems.
Joint Commission RUAIH and Joint Commission–CHAI materials are used as international organizational reference points. RUAIH is currently a voluntary U.S. certification program and does not create legal requirements for healthcare organizations in Viet Nam.
The Healthcare AI Leadership Oversight Dashboard and Governing Body and Executive AI Oversight Framework proposed in this article are practical organizational frameworks. They are not Vietnamese statutory frameworks, Joint Commission standards or CHAI templates.
Operational delegation of AI governance does not itself alter statutory responsibilities, professional obligations or legal liability. Applicable responsibilities depend on the organization, AI system, legal role, intended use and deployment context.
References
National Assembly of Viet Nam. Law No. 134/2025/QH15 on Artificial Intelligence. Adopted 10 December 2025; effective 1 March 2026.
Government of Viet Nam. Decree No. 142/2026/NĐ-CP Detailing a Number of Articles and Measures for Implementation of the Law on Artificial Intelligence. Issued 30 April 2026; effective 1 May 2026.
Prime Minister of Viet Nam. Decision No. 33/2026/QĐ-TTg Promulgating the List of High-Risk Artificial Intelligence Systems. Issued 30 June 2026; effective 15 August 2026.
National Assembly of Viet Nam. Law No. 15/2023/QH15 on Medical Examination and Treatment, as amended; current consolidated text: Consolidated Document No. 26/VBHN-VPQH dated 26 February 2026. Law effective from 1 January 2024.
Ministry of Health of Viet Nam. Circular No. 19/2013/TT-BYT Providing Guidance on Management of Quality of Medical Examination and Treatment Services in Hospitals. Issued 12 July 2013; effective 15 September 2013; currently recorded as in force.
Joint Commission. Responsible Use of AI in Healthcare Certification. Launched 1 June 2026.
Coalition for Health AI. AI Governance Playbooks. Released 27 May 2026.
Joint Commission and Coalition for Health AI (CHAI). Guidance on the Responsible Use of AI in Healthcare (RUAIH). Initial high-level guidance, released 17 September 2025.