Boards have spent at least two years learning to fear what AI might do inside their organisations: the shadow tools used without permission, the models that reach decisions no one can quite explain, the compliance exposure that follows close behind. Far less attention has gone to the opposite scenario: a workforce that cannot use AI safely or well because no one has equipped it to.

The latest international evidence puts a number on that blind spot. In a survey of 6,000 employees and HR leaders across 11 countries in Europe, Asia and Latin America, 68 per cent of HR directors named AI and automation the single biggest driver of skills change over the next two years. Only 32 per cent of employees said their employer had trained them for it. That is not a training shortfall. It is a control gap, and it sits on the part of the organisation that most boards never audit.

A liability that never reaches the balance sheet

Skills do not depreciate on a timetable a board can read. They erode quietly, until the morning a strategy turns out to depend on a capability the workforce no longer has. The World Economic Forum estimates that 39 per cent of workers’ core skills will be transformed or outdated by 2030. The barometer points the same way: nearly one in four jobs is exposed to skills obsolescence within three years, and 41 per cent of employees already say training reaches them too late to be useful.

The asymmetry is hard to miss. Financial capital is reviewed every quarter, with audited figures and a committee paid to interrogate them. Human capital, usually the largest cost in the business and the thing every strategy actually runs on, is handled through a narrative paragraph and a training budget no one stress-tests. As the distance between what the technology can do and what employees can do with it widens, the board is generally the last to find out.

Why this is a board matter, not an HR one

There is something faintly self-serving in a chief people officer arguing that workforce capability belongs on the board’s agenda. The CHRO, the CPO, the HR director—whatever name the role currently travels under—has spent a decade asking for exactly that seat. So discount the messenger if you like; the argument stands on its own.

Two forces are turning workforce capability from an operational concern into a governance one. The first is risk. An organisation that deploys AI faster than it builds the competence to govern it is quietly accumulating exposure: decisions taken with tools few understand, errors that scale at machine speed, and growing dependence on the handful of people who do understand them.

Boards already accept that logic for cybersecurity and for financial controls. Capability is a control, and an uncontrolled one is a liability. The WEF’s employers are blunt about the stakes: 63 per cent call skills gaps the single biggest barrier to transforming their business by 2030. A board that cannot see its own skills gap cannot see one of its largest obstacles to executing strategy.

The second force is disclosure and, for a UK board, it is closer than it looks. The UK Corporate Governance Code already expects the board to understand its workforce, engage with it and take ownership of culture, rather than delegate the lot to management. Yet workforce reporting in FTSE annual reports remains thin and largely narrative.

Skills, capability and the readiness of people to absorb new technology are rarely measured, rarely owned at board level and rarely tested by anyone. A UK board that can describe its AI strategy in detail but cannot say how many of its people are equipped to deliver it is carrying a control gap under a comply-or-explain regime that will eventually ask about it.

And the direction of travel runs one way. Across the Channel, the European Union has already turned the principle into hard reporting: the standard on own workforce, ESRS S1 under the Corporate Sustainability Reporting Directive, reaches explicitly into training and skills development, and although the 2026 Omnibus reforms narrowed who must report, they lowered the regulatory floor, not the market’s expectations.

The United States has required listed companies to address human capital since 2020, and investors and lenders ask for workforce data whether or not a code or a directive compels it. What is comply-or-explain in London today is becoming mandatory disclosure elsewhere. The exposure is the oldest one in governance: saying something in the annual report that the organisation does not, in practice, govern.

What boards should be asking

None of this asks the board to become a learning and development department. It asks the board to treat workforce capability as the risk it already is. A few questions are enough to begin:

• Is skills obsolescence on the risk register, with an owner and a measure, or is it buried in an HR update?
• Do we track the gap between the AI we are deploying and the capability of the people expected to use it?
• Is our workforce disclosure backed by data and controls, or is it narrative?
• Who on this board owns workforce capability, and when did we last discuss it as a strategic risk rather than a cost line?

Governing AI by watching the technology is an understandable instinct, and it looks at the wrong half of the problem. The tools are moving faster than the people expected to use them, and the space between the two is where value leaks and risk gathers. Boards that price that gap now, in the UK or across the EU, before regulators, investors and their own employees press them to, will be governing the transformation rather than reporting on it once it has already happened.

Alessandro Reati is a work psychologist and head of people and culture at learning and development provider Cegos Italia.