AI has moved from the management meeting to the boardroom. Directors are now expected to oversee a technology many have never used hands-on, and the temptation is either to defer to the experts or to wait for the picture to settle. Neither is oversight. The good news is that boards do not need to become technologists. They need to do what effective boards have always done — ask simple, persistent questions and expect straight answers.

Why boards can’t sit this one out

AI decisions now carry strategic and fiduciary weight at the same time. Used well, AI compounds: organisations that learn faster pull ahead on cost, service and speed. Used carelessly, it embeds privacy, security and reputational risk deep inside everyday operations — often invisibly. And doing nothing is itself a decision. Staff are already using AI tools, sanctioned or not, so silence at the top does not mean the organisation has no AI position; it means the position is being set by default, one workaround at a time.

Five questions that cut through

• Where is AI already being used in our organisation — including the tools nobody approved?
• Who is accountable when an AI-assisted decision goes wrong, and how would we know it had?
• What does responsible use mean for us, and is it written down somewhere staff can actually find?
• What are our AI investments returning, and what have we stopped doing as a result?
• How are we, as directors, building our own fluency rather than outsourcing our judgment?

The pragmatic takeaway

A standing agenda item beats an annual deep-dive. Start small: an inventory of current AI use, a one-page statement of what responsible use means in your context, and a regular report against both. Boards that ask these questions consistently do more than manage risk — they signal to the whole organisation that AI is core business, not an experiment happening at the edges.