AI Trust, Assurance, and Governance (AITAG) Certificate Program

Virtual Modules: 1–5 p.m. EDT

• January 2027 dates will be announced soon

Please note that we use Zoom for all program deliveries.

• The deadline for Cohort 1 will be announced soon. Please submit your application as soon as possible for consideration.

• If space remains, we will continue to accept applications on a first-come, first-served basis.

• $4,250 for the entire program
• $3,750 discounted rate for Carnegie Mellon alumni — including our Chief Information Officer (CIO), Chief Information Security Officer (CISO), Chief Risk Officer (CRO), Chief Data and Artificial Intelligence Officer (CDAIO), Chief Information and Digital Officer (CIDO) and Chief Digital Officer (CDigitalO) programs — U.S. government employees, veterans and employees of nonprofit organizations
• Program cost is a flat rate with no additional fees

Please note: Due to the non-credit-bearing nature of the AI Trust, Assurance, and Governance (AITAG) Certificate Program, students are unable to apply for tuition assistance, scholarships or VA benefits. Program costs cannot be itemized.

The AI Trust, Assurance, and Governance Certificate Program is designed for professionals responsible for deploying, securing and governing AI systems, including:

• Chief AI Officers, Chief Data Officers, Chief Information Officers, Chief Information Security Officers and Chief Technology Officers
• AI engineers, machine learning engineers, software architects and AI solution architects
• AI product managers, program managers and technology innovation leaders
• AI governance, risk, compliance and responsible AI practitioners
• Cybersecurity professionals responsible for securing AI systems and infrastructure
• Model validation, testing, quality assurance and internal audit professionals
• Enterprise architects, digital transformation leaders and technology strategy professionals
• Data governance, machine learning operations (MLOps), LLM operations (LLMOps) and AI platform leaders
• Government, defense and public sector professionals responsible for deploying and assuring mission-critical AI systems

Through a structured, four-module experience, participants will develop the capabilities to:

• Explain why AI requires different assurance approaches than traditional software and apply leading frameworks to enterprise deployment
• Design and execute structured AI evaluation methodologies, including red teaming and adversarial testing
• Assess AI security architecture and build layered defenses for LLM and agentic deployments
• Design governance models, establish operational controls, and implement continuous monitoring and assurance processes
• Define executive accountability for AI risk and operationalize assurance across the AI lifecycle

Carnegie Mellon University is a global leader in artificial intelligence, machine learning and cybersecurity. Heinz College combines this technical depth with a longstanding focus on policy, management and real-world impact — and the AITAG program is developed in partnership with AIMSEC, CMU's interdisciplinary center for AI measurement science and engineering.

This program reflects CMU's distinctive approach: equipping leaders to understand, validate, secure and govern AI responsibly and at scale. Learn from CMU faculty at the forefront of AI research and practice.

Earning your executive education certificate is just the beginning of a lifelong relationship with CMU; we're here to help you advance your career throughout your professional life. The Heinz College Executive Advantage program is meant for people who always want to be ahead of the curve — those who want to lead the conversation, not just be a part of it.

Executive Advantage was designed with you, C-suite executives and lifelong learners, in mind. In addition to the valuable skills you gain through our executive certificate programs, you now have access to workshops, leadership summits and conferences.

What Must Be Assured — and Why AI Changes Enterprise Assurance

This opening module establishes the conceptual foundations of AI assurance and explains why modern AI systems require fundamentally different assurance approaches than traditional software. Participants develop a common language for AI assurance by examining how generative and agentic AI differ from conventional applications, why evaluation must occur across multiple levels, and how organizations should define acceptable levels of risk before deployment. The module introduces leading AI assurance frameworks and governance models that provide the foundation for the remainder of the program.

Key topics include: why AI differs from deterministic software; the AI assurance landscape (performance, security, safety, fairness, privacy and accountability); model, system and application-level assurance; the AI Assurance Lifecycle; and AI assurance standards and frameworks including the National Institute of Standards and Technology (NIST) AI Risk Management Framework (RMF), ISO/IEC 42001, and emerging regulatory expectations.

How Do We Know AI Systems Perform As Intended?

This module examines the methods organizations use to evaluate AI systems before deployment. Participants learn the strengths and limitations of benchmark testing, adversarial evaluation and real-world validation, while gaining practical insight into why no single testing approach provides sufficient evidence for deployment decisions. Particular attention is given to agentic AI, where autonomous behavior introduces new testing challenges beyond traditional generative AI.

Key topics include: structured AI evaluation (intrinsic model evaluation, adversarial testing, field testing); evaluating AI performance (reliability, hallucinations, robustness, bias, toxicity, correctness); adversarial testing and red teaming (prompt injection, jailbreak testing, retrieval-augmented generation (RAG) manipulation, hybrid human-AI testing); and testing agentic AI (multi-step reasoning failures, tool use, memory, agent interactions, reward hacking, specification gaming).

How Can AI Systems Be Compromised?

This module focuses on defending AI systems against adversarial attack. Participants examine how the attack surface changes when organizations deploy large language models and autonomous agents, and how AI-specific security risks differ from traditional cybersecurity. The emphasis shifts from testing system behavior to protecting production AI environments through secure architectures, resilient design and AI-specific security controls.

Key topics include: the AI threat landscape (AI-specific attack vectors, threat actors, foundation model risk); securing AI architectures (model, data, API, and identity and access management, third-party AI risk); securing agentic systems (tool access, memory manipulation, agent communication, retrieval security, autonomous execution controls); and building resilient AI systems (defense-in-depth, least privilege, architectural resilience, AI-specific incident response, recovery planning).

How Do Organizations Operate Trustworthy AI At Scale?

The final module focuses on sustaining trustworthy AI throughout production operations. Participants examine how organizations operationalize AI assurance through governance, continuous monitoring, human oversight, accountability and enterprise risk management. Rather than emphasizing technical testing, the discussion centers on the organizational capabilities required to safely scale AI while maintaining executive oversight and regulatory confidence.

Key topics include: operational AI governance (governance structures, roles and responsibilities, executive oversight, board reporting); operational controls (human oversight strategies, decision rights, escalation mechanisms, operational safeguards); continuous assurance (runtime monitoring, model drift, performance degradation, key performance indicators (KPIs) and key risk indicators (KRIs), incident management); and enterprise accountability (cross-functional governance, auditability, documentation, regulatory compliance, continuous improvement).