Boards and directors: questions about AI

AnsweredWhat should a board ask about AI?#

Twelve questions, each paired with the answer that should worry you. Ask two or three in passing rather than as an agenda item.

AnsweredWhat is the difference between AI governance and AI leadership?#

Governance checks decisions that have been made. Leadership makes them. Most organisations built the first and are waiting for it to produce the second.

AnsweredDo AI agents in companies act without real-time human oversight?#

Mostly, on the executives' own account: 85 per cent of senior AI decision-makers at large US listed companies using agentic AI told EY in mid-2026 that at least some systems execute actions without real-time human involvement, and 26 per cent could not detect an unauthorised agent. Self-report from 202 people, and 'no real-time involvement' is not the same as 'cannot be stopped'.

AnsweredWhat should management report to the board about AI every quarter?#

Answered. The declared augmentation-or-replacement position annually, the capability floor with an owner and a date, decision rights in writing, reversal thresholds, and one reconstructed decision. If the paper reports adoption rates and hours saved, the board is being shown procurement.

AnsweredHow does the board know management's claims about AI are true?#

Open, and the assurance question underneath most of this territory. Every other material claim to a board has an independent check behind it. AI claims mostly do not, and the people best placed to verify them report to the people making them.

AnsweredDoes the board need independent assurance over AI?#

Open. Internal audit is the obvious home and is rarely resourced for it. The harder problem is that assuring an AI system requires the competence to evaluate it, which is scarce in exactly the third line.

AnsweredWhat should the board do about the September 2026 AI safety warnings?#

Five decisions that hold whatever the odds: which decisions machines may make in the company's name, who can stop each system, what people must remain able to do unaided, how the board would find out, and what management's assurances rest on.

AnsweredHow many AI decisions an hour is our human oversight actually checking?#

Decisions per hour, times the minutes a real check takes, against the attention available. Where the first exceeds the second the approval is a sample, and the board should know the sampling rate and who owns the decisions that were not checked.

AnsweredWhich AI failures must be reported to the board?#

Open, and the reporting threshold has a perverse property: the failures worth escalating are the quiet, systematic ones, and the ones that get escalated are the loud, visible ones.

AnsweredShould AI oversight sit with the full board or a committee?#

Open, and contested. A committee gets depth and creates the impression the rest of the board need not engage. The full board gets engagement without the time to go deep. The audit-committee default imports a compliance frame onto a strategy question.

AnsweredDoes our board have enough AI expertise?#

Open, and usually the wrong question. What a board needs is enough understanding to tell a real control from a described one, which is not the same as technical depth and is not obtained by appointing one specialist.

AnsweredShould we appoint a director with AI expertise?#

Open, and the cyber precedent is not encouraging. Appointing a single expert tends to concentrate rather than raise competence, and the rest of the board defers on exactly the questions it should be asking.

AnsweredHow should directors themselves use AI?#

Answered, and the answer is yes but not for productivity. A director who has never watched a model produce a confident, wrong answer in their own domain has no calibration for the risk they are overseeing. Note the measurement gap: PwC has 35% of directors self-reporting AI in oversight, while Deloitte's governance professionals were largely unsure whether their directors use it at all.

Better answered elsewhere Measured board practice rather than advice about it, including how much of it is happening without board-specific policy. · Board Practices Quarterly: How boards are using AI today

AnsweredShould directors put confidential board papers into AI systems?#

Open, and the one question here with an immediate practical answer: not into public tools. PwC's guidance says so directly. What remains open is the enterprise case, where confidentiality is contractual and the residual question is privilege and discoverability.

AnsweredCan a director rely on an AI-generated summary of a board paper?#

Open, and the sharpest question in the cluster. Ayinde established that the duty to verify does not transfer to whoever used the tool, and that it travels upward. Applying that to a director who read a summary rather than the paper is untested and uncomfortable. The Deloitte survey found summarising board materials is one of the most common uses.

AnsweredHow should a board govern AI agents?#

By requiring five things rather than approving agents one by one: an inventory, a named owner for each agent, written decision limits, a tested stop, and supervisors able to judge the work, with evidence that they hold. Autumn 2026 surveys by EY, KPMG and BCG show deployment running ahead of these controls.

AnsweredCould our directors be liable for overstating what our AI does?#

No UK case yet, but the routes exist: issuer liability under Schedule 10A to FSMA for knowing or reckless misstatements, UK MAR where a person ought to have known, and section 463 of the Companies Act for the strategic report. In the US, AI-related securities class actions in the first half of 2026 nearly matched all of 2025.

AnsweredWhat should a board ask about AI and cyber risk?#

How AI has changed the threat and what management has done differently; whether a payment requested by a deepfaked executive would be stopped; what the company's own AI tools can reach if manipulated; how much AI use happens outside approved tools; and when the board last rehearsed an AI-enabled attack.

AnsweredWhat should a chair ask the chief executive about AI?#

Questions that get an answer rather than a reassurance, and what a good answer sounds like when it arrives.

AnsweredHow does a chair test an AI answer without being technical?#

By asking which claims can be checked and how, rather than by understanding the system.

AnsweredWhat does a non-executive director need to know about AI?#

Less than is usually assumed, and rarely technical. The gap is knowing which management claims can be verified.

AnsweredWhat should a new non-executive ask in their first board meeting about AI?#

Twelve questions that establish capability, evidence, accountability and pipeline rather than adoption.

AnsweredWhat should a company secretary put in a board paper about AI?#

What the board needs each quarter, agreed with management so it arrives without being chased.

AnsweredWhat does the UK Corporate Governance Code require a board to say about AI?#

The Code sets what a board must be able to declare. The question is whether the declaration survives somebody asking how you know.

AnsweredWhich senior manager is accountable for AI under the Senior Managers Regime?#

Accountability attaches to a person, and AI makes it hard to say which person.

AnsweredCould using AI change what a director is personally liable for?#

Duties do not change. What changes is how hard it becomes to show they were discharged.

AnsweredAI is on our board agenda for the first time. Where do we start?#

With what the board has to be able to establish, rather than with a briefing on the technology.

AnsweredOur board evaluation is coming up. Should AI be in it?#

Readiness is a claim, and a claim a board makes about itself is the one most worth testing.

AnsweredAn investor has asked what our board is doing about AI. What do we say?#

What was considered, who can stop a system, and what capability the organisation has kept.

AnsweredSomething went wrong with an AI system. What does the board need to establish?#

Whether it was reportable, who could have stopped it, and whether anybody did.

AnsweredManagement wants to scale an AI pilot. What should the board require first?#

Evidence that would trigger reversal, agreed before the deployment rather than after it.

AnsweredA supplier says their product is AI-powered. What should the board ask?#

Who verified it, against what source, and who bears the consequences if it is wrong.

AnsweredWho can stop an AI system in our organisation?#

Usually nobody has been named, and the question is rarely asked until it matters.

AnsweredDoes a board need an AI kill switch?#

A stop control nobody has used is a control nobody knows works.

AnsweredHow does a board know its people can still do the work without the system?#

By assessing capability rather than output, which is the measure that stops telling you anything first.

AnsweredWhere are our next senior people coming from if AI does the junior work?#

A succession question that arrives dressed as a technology question.

AnsweredWhat happens to institutional memory when AI does the writing?#

What an organisation knows lives in the people who wrote it down, until it does not.

AnsweredWhat does AI literacy mean for a board?#

Enough to read the paper and know which sentence to doubt, which is not the same as understanding the model.

AnsweredShould the whole board be trained on AI, or is one expert director enough?#

One expert director can become the person everybody else defers to, which is the opposite of oversight.

AnsweredWhich decisions should a machine never make here?#

The boundary is an organisational decision, and most organisations have never written theirs down.

AnsweredWho decides which decisions AI is allowed to make?#

Decision rights get allocated by default when nobody allocates them on purpose.

AnsweredIs human in the loop enough?#

A person who approves faster than they can evaluate is a record of approval, not a safeguard.

PartialWhat should a board ask when an AI strategy is presented?#

Ask what it assumes people will still be able to do, and how that will be checked. Most strategies do not answer either.

PartialWhat evidence should a board demand before scaling an AI pilot?#

Twelve questions, each paired with the answer that should worry you. Time saved is not on the list.

PartialWhich AI decisions should come to the board?#

Partly answered. The defensible line is not size of spend but reversibility and whether the decision changes what the organisation can do unaided. Most escalation criteria use spend, which is the wrong axis.

PartialWhat would an effective board AI dashboard contain?#

Partly answered, with a caution. The agenda items are known; turning them into a dashboard risks converting the one question that resists metrics, whether anybody got better at anything, into a number that does not measure it.

PartialHow should AI use in board decision-making be recorded?#

Partial from 14 September 2026. There is a concept for the record and no board practice built on it: decision provenance, proposed by Singh, Cobbe and Norval in 2019, logs the inputs, the decision and its downstream effects without attempting to explain any reasoning. Minutes record decisions and dissent, not the provenance of a director's understanding. Whether that needs to change is the question, and the answer has litigation consequences nobody wants to be first to discover.

PartialDoes our board have the AI skills to oversee it, and how would we know?#

PwC's survey of 561 board members, reported by Accounting Today on 23 September 2026, found 71 per cent saying boards need stronger AI skills for oversight and 35 per cent fearing that over-reliance on AI output weakens human judgement. Directors self-reporting a deficit is a claim about mood; the test on the oversight page is whether the board can answer the five questions without management's help.

PartialHow many boards have an AI governance framework?#

Two per cent, on Institute of Directors New Zealand data quoted by B2B News on 30 September 2026, against 79 per cent of leaders using AI weekly; the method is not given in the article. The figure is one country's and unverified here. The board oversight page holds what a framework has to contain to be worth counting.

PartialCan a board delegate accountability for an AI decision?#

No, on the paper two European bodies sent boards on 7 October 2026: Accountancy Europe and ecoDa write that accountability for AI outcomes cannot be delegated to technology and stays with management and ultimately the board, whatever the degree of automation. It is advice and not law. The oversight page holds what a board has to see to discharge it.

PartialHow should a board minute a discussion about AI?#

So that it records what was considered and what was decided, not that a presentation was received.

OpenCould using AI change a director's duties?#

Open, and a question for counsel rather than for this research. The duty to exercise independent judgement and reasonable care is jurisdictional. What can be said is that Ayinde made a professional verification duty non-delegable and upward-travelling, and no reported case yet applies that reasoning to directors.

OpenHow should the board oversee AI used by suppliers and third parties?#

Open, and the gap most likely to produce the first governance failure. Third-party risk processes ask whether a supplier has a policy. They do not ask whether the supplier's AI use is eroding the capability you are buying from them.

OpenWhat should the board do when staff warn that an AI system is not being monitored?#

The New York Times reported on 29 September 2026, in a story read in syndication, that two OpenAI employees warned executives months before the agent incidents that new models were not being appropriately monitored in testing and were told the tests had to move quickly. No study measures what makes such a warning travel. A board can settle it for itself: count the warnings that reached it last year and what happened to each.

OpenWhat should a charity board ask about AI?#

Trustee boards carry the same duties with less support, and almost nothing is written for them.

OpenWhat should school governors ask about AI?#

Governors are being asked to approve AI in schools with no governance material written for the role.