Beyond the AI Strategy: The Risks Your Board is Missing

Beyond the AI Strategy: The Risks Your Board Is Missing

Most organizations are now developing an AI strategy.

Many have approved pilot programs, invested in new tools, and begun exploring how artificial intelligence can improve productivity, customer experience, and business performance.

But approving an AI strategy is not the same as understanding the risks behind it.

In this episode of Three Takes on AI, Brian Silverman, Campbell Robertson, and Michael Muhlfelder begin with a list of 12 questions every board should be asking about AI. Those questions cover issues ranging from vendor dependency and regulatory exposure to confidential data, accountability, workforce disruption, and whether board members are sufficiently AI-literate to challenge management.

The discussion quickly moves beyond the original list.

What happens when a mission-critical AI application depends on a model provider that changes its terms, raises its prices, introduces a new model, is acquired, or disappears altogether?

Moving from one AI model to another may sound straightforward, but applications are often tightly connected to the behavior, capabilities, and economics of a specific model. Vendor flexibility cannot simply be assumed.

The hosts also examine the limits of policies and governance checklists. An organization may have assigned responsibility, added AI to its risk register, and documented acceptable-use policies. But those measures mean little without real operational controls. When an AI agent acts outside its intended boundaries, a model produces a damaging result, or confidential information is misused, someone must have the ability, and the authority, to stop the system.

That raises a larger question: who has the power to challenge or veto a bad AI decision?

The conversation explores whether boards need broader input from employees, customers, governance leaders, outside experts, and other stakeholders who may see consequences that executive leadership does not. AI decisions can affect far more than technology. They can reshape jobs, judgment, organizational authority, workplace culture, and relationships with customers and society.

The human role also requires more attention. Putting a “human in the loop” does not automatically create effective oversight. If that person is poorly trained, does not understand the decision being reviewed, or simply approves what the system recommends, the control may exist only on paper. The problem is not merely human error. It may reflect poor system design, weak training, unclear accountability, and a failure to define what meaningful human review should accomplish.

Ultimately, the episode asks whether organizations need a dedicated AI board or oversight body composed of business, technology, governance, legal, workforce, and external perspectives. Such a group could evaluate proposed AI initiatives, review existing deployments, challenge assumptions, and make certain that the organization has meaningful guardrails before problems occur.

AI cannot be treated as just another technology implementation. Its impact on business economics, employees, decision-making, risk, and organizational power makes it a business and governance issue from the beginning.

The train wreck may still be preventable, but only if boards begin asking harder questions before AI decisions become irreversible.