AI adoption is rising across finance and accounting, but adoption alone does not guarantee results. Finance teams are still working through fundamental questions: How can organizations trust AI with mission-critical financial processes? What governance is needed as AI becomes embedded in finance? And is human review enough to control the risks created by increasingly automated workflows?
In this Vision 2026 session, Mark McDonald, CEO and Founder of Finance Next, examines why many finance organizations struggle to translate AI adoption into business value. Drawing on his experience in finance leadership, software engineering, AI, and advising finance organizations at Gartner, McDonald identifies three major hurdles: a lack of strategic direction, low trust in AI, and new security and governance vulnerabilities.
McDonald argues that AI governance should enable adoption rather than simply restrict it. Finance teams can create greater confidence by integrating AI into their broader finance strategy, establishing clear accountability for AI-enabled processes, classifying use cases by risk, maintaining an AI register, addressing shadow AI, and applying appropriate access controls.
The session also challenges one of the most common approaches to AI governance: relying on a human-in-the-loop to review AI output. McDonald explains how manual review can introduce scalability problems and a false sense of control as AI-driven transaction volumes increase. Instead of depending on a single human backstop, he proposes building multiple layers of automated validation and independent review into AI-enabled financial processes.
Those layers can include data and configuration standards, automated validation, exception handling, fraud and security checks, transaction logging, and independent error and anomaly detection. The objective is not to assume AI is error-free, but to create a repeatable governance structure that can demonstrate how reliability is being established and monitored.
McDonald also explores the importance of independence in AI governance. When one AI system executes a financial process, a separate process can independently evaluate its output for errors or anomalies. This extends familiar finance concepts such as segregation of duties and independent review into an increasingly automated environment.
Looking further ahead, the session examines how AI itself could help automate finance governance. By documenting governance rules, responsibilities, controls, policies, and monitoring requirements in a structured framework, organizations can create a foundation that AI can eventually help query, maintain, and operationalize. Potential applications range from drafting policies and evaluating processes against governance requirements to designing controls and documenting their execution.
For finance leaders deciding where to begin, McDonald recommends connecting AI initiatives to a three-year finance roadmap, stratifying AI use cases by risk, establishing an AI register, assigning accountability to process owners, addressing shadow AI, and documenting governance requirements.
The central takeaway: effective AI governance is not simply about slowing AI down or adding more manual review. It is about creating the structure, accountability, independent validation, and transparency finance teams need to adopt AI with greater confidence.
Chapters
02:48 Why AI adoption isn’t translating into success
05:04 AI adoption vs. AI success
06:10 Three hurdles holding finance AI back
09:09 The AI trust problem
11:59 Why human review can fall short
14:07 AI security, prompt injection, and shadow AI
16:30 Using governance to overcome AI hurdles
18:04 How governance can accelerate AI adoption
20:37 Building a three-year finance AI strategy
23:59 Building trust through process ownership and transparency
26:05 Stratifying AI use cases by risk
27:40 Independence and segregation of AI duties
29:23 AI access controls, shadow AI, and vendor governance
34:19 Rethinking human-in-the-loop review
36:16 Building automated layers of AI validation
39:37 Exception handling, fraud, and security
41:31 AI transaction logs and AI registers
43:13 Independent error and anomaly detection
48:13 Using AI to automate governance
49:10 Building a governance master framework
51:59 Key AI governance takeaways
54:27 Q&A
56:33 Where should finance teams start?
