Boards adopt AI but nobody told legal

This spring, 82% of American public-company directors used generative AI in their board work, up from 66% last September. Only 6% sat on a board that had written a rule for it. That gap, between what directors already do and what their boards have decided they may do, is the story of AI in the boardroom this year. The Diligent Institute, which surveyed 104 directors in June, found that 54% have no guidance at all for how a director should use the tool. The people charged with oversight are improvising.

This is not a boardroom quirk; it is the shape of the whole enterprise. IBM's Institute for Business Value, surveying 2,000 technology chiefs across 33 countries, found 77% saying AI adoption has already outrun their ability to govern it. The gap directors feel in their own seats is the gap their companies feel on the floor.

Boards are not ignoring AI. The Conference Board reports that the share of S&P 500 companies naming AI as a risk factor jumped from 12% to 83% in two years. What sits behind the disclosure is thinner. Only 2.7% of directors disclose any AI expertise, and just 9% of executives call their company very prepared to comply with AI regulation. Diligent's wider survey of directors puts a figure on the discomfort: 8% rate their board strong on AI, the lowest score of any skill measured, and 22% have a formal AI governance policy. Deloitte, surveying about 700 directors and executives across 56 countries, finds two-thirds say their board has limited to no knowledge of AI, and 31% say it is not yet on the board agenda at all.

So boards are building the seat as they sit in it. ISS-Corporate finds the share of S&P 500 directors with disclosed AI expertise nearly doubled in two years, from 11% to 20%, while a third of the index now discloses some board oversight of AI. Among the Fortune 100 the shift is sharper still: EY reports that boards assigning a committee to oversee AI rose from 11% to 40% in a single proxy cycle, and AI now appears as a risk factor in 89% of their annual filings. The qualification of an AI-literate director is being written into committee charters this year. That is a rare opening, and a precarious one.

Here the story becomes Athena's. The fastest-growing route onto a board is the skill our members already build. Spencer Stuart's 2025 Board Index finds technology is now the single most common background among newly appointed women directors, 22% of their S&P 500 appointments, up from 17% a year earlier. The credential boards are short of is the one the AI Academy issues.

The opening is real, and so is the risk of missing it. The same data show the door narrowing. Egon Zehnder's Global Board Diversity Tracker finds women's share of new board appointments worldwide fell to 14.2%, down from 17.2% in 2020. In America, women are 35% of S&P 500 directors but only 38% of new ones, a share that has slipped for three years running. 50/50 Women on Boards finds the Russell 3000 added just 0.1 of a percentage point of women last year, and that women of color went backwards, to 7.4% of seats. More demand for the skill, fewer of the seats.

One assumption is worth retiring on the way out. The reflex that women lean away from AI does not hold at the top. AlixPartners, surveying 3,000 senior executives, finds women leaders more optimistic about the technology than men, 36% extremely optimistic against 30%. What holds women back in the boardroom is not confidence in the tool, but access to the seat where it is now governed.

Most of these counts come from what companies disclose in proxy statements and what directors tell surveyors, not from an audit of what a board can actually do. A committee named in a charter is not the same as competence in the room, and the AlixPartners figure measures executives, not directors alone. The direction, across nine separate studies, is hard to miss.