The board's AI blind spot: governance without understanding.

There is a growing expectation — from regulators, investors, and the public — that boards will govern AI responsibly. This expectation is reasonable, and in many respects overdue. AI systems are increasingly making or informing decisions that affect customers, employees, and wider society. Boards that are not engaged with how these systems work, what they are used for, and what oversight exists, are leaving a significant gap in their governance responsibilities.

The problem is that most boards are not yet equipped to meet this expectation. Not because board members are unintelligent or uninterested — they are neither — but because they have not been given the context, the frameworks, or the access to information that they would need to govern AI with genuine effectiveness.

What boards are currently doing

In most organisations, board-level engagement with AI takes one of two forms. Either AI appears occasionally as a standing item on the technology or audit committee agenda — typically presented by the CTO or CIO in technical terms that most board members find difficult to interrogate — or it is treated as a subset of the wider digital strategy, with no specific governance lens applied to it.

Neither approach is adequate. The first produces passive acknowledgement rather than active governance. The second treats AI as a capability question when it is also, increasingly, a risk and accountability question. What is needed is a more structured approach that gives boards the tools and the information to exercise genuine oversight.

The four questions every board should be asking

Effective AI governance at board level does not require boards to become data scientists. It requires them to ask the right questions — consistently, and with enough fluency to evaluate the answers they receive.

• Where is AI being used in our organisation, and what decisions is it informing or making? Many boards do not have a complete inventory of AI deployments. Without this, oversight is impossible.
• What is our accountability structure for AI? When an AI system makes a decision that causes harm — to a customer, an employee, a member of the public — who is accountable? The answer should not be "it's complicated".
• How are we identifying and managing AI risks? This includes technical risks (model failure, bias, data quality) but also reputational, regulatory, and ethical risks. The board should be receiving regular, plain-language reporting on these.
• Are we keeping pace with the regulatory environment? AI regulation is evolving rapidly. The EU AI Act, incoming UK frameworks, and sector-specific guidance all have implications for how organisations need to govern and document their AI systems.

Building board capability

The starting point for most boards is education — not technical education, but governance-focused education that helps board members understand the specific risks and accountability questions they need to engage with. This is one of the fastest-growing areas of our executive education work, and we consistently find that well-designed, practically-focused sessions produce a step-change in the quality and confidence of board-level AI conversations.

The next step is structural: putting in place the reporting, the committees, and the escalation mechanisms that give boards a regular, clear line of sight into AI governance across the organisation. This is not complex to design, but it does require the organisation's executive team to take the board's needs seriously and to invest in making AI visible at the right level of detail.

---

DeepSlate works with boards and executive teams on AI governance frameworks and board-level AI education. See our AI Governance Framework product or get in touch.