Board oversight of AI: the questions directors should ask
A board does not need to understand how a model works. It needs to know where the company uses AI, what could go wrong, who is accountable and whether the investment is paying back.
Board oversight of AI means directors satisfy themselves that management has a strategy for AI, understands and controls the risks, and reports honestly on results. In practice the board asks a standing set of questions across six areas, sees a short scorecard each quarter, and is told promptly about serious incidents. Oversight is asking and testing, not managing the program.
NIST
ISO
European Commission
OECD
Anthropic
Microsoft 365
Where the numbers come from
These counts come straight from the lists on this page. Nothing here is a survey result or a client outcome.
Which committee should own AI oversight?
The full board owns strategy. Risk oversight usually sits with the committee that already covers technology and cyber risk, often audit or risk. What matters is that it is written into a charter rather than assumed.
- Put AI on the full board agenda at least once a year for strategy.
- Give one committee the standing risk review.
- Record who briefs the board and how often.
What should the quarterly scorecard show?
One page: where AI is in use, what it has returned against target, what it has cost, what went wrong and what decisions are coming.
- Use cases live, in pilot and stopped.
- Results against the target set when each was funded.
- Spend against budget, split into one-time and running costs.
- Incidents and near misses, with what changed afterwards.
- Regulatory changes that affect the company.
How much should directors know about AI?
Enough to ask good questions and to tell a real answer from a confident one. That means a shared briefing on how the tools work, where they fail and what the main frameworks expect.
- A working session with hands on the tools is worth more than a slide deck.
- Bring in outside expertise for the areas the board lacks.
- Revisit yearly; the technology and the rules both move.
The EU AI Act has required organisations that provide or deploy AI systems in the EU to ensure a sufficient level of AI literacy among their staff since 2 February 2025. Directors of companies with EU exposure should ask how that is met.
Questions executives ask about this
Is AI oversight a legal duty for directors?
Directors' duties of care and oversight apply to significant risks whatever their source, and AI can be one. What that requires depends on your jurisdiction and your company; ask your counsel. This page is general guidance, not legal advice.
Should the board approve individual AI projects?
Not normally. The board approves the strategy, the risk appetite and material investments, and management runs the portfolio within those limits.
How often should AI be on the agenda?
Strategy at least yearly, the scorecard quarterly, and serious incidents as they happen.
Do small private companies need this?
A lighter version, yes. An owner-led company with an advisory board still benefits from asking the same questions once or twice a year.
Primary sources to check against
AI Risk Management Framework, AI RMF 1.0 (National Institute of Standards and Technology)
ISO/IEC 42001:2023, AI management systems (International Organization for Standardization)
AI Act: the regulatory framework for AI and its application timeline (European Commission)
OECD AI Principles (OECD.AI Policy Observatory)
Frameworks, laws and product details change. Confirm the current text with each publisher and with your own counsel. Consultants in a Box is an independent consultancy; this page is not published or endorsed by any organisation named. Names and logos are trademarks of their owners.
Related guides
AI governance frameworks
You do not need to adopt every framework. You need to know which ones are voluntary, which one is law where you operate, and how to build one set of controls that satisfies them.
Read the guideChief AI officer
Someone has to own AI across the company. That does not have to be a new full-time executive. The decision is about accountability first and the title second.
Read the guideAI for CFOs
The build is the small number. Licences, usage, integration upkeep, review time and training recur every month, and a business case that leaves them out will look better than the result.
Read the guideAll AI for executives guides
The full leadership playbook, from readiness to agents.
See the hubWant to talk this through for your company?
Book a call and we will go through where you are, what to decide first and who should own it. Or start with the free business analysis.
