The 2.7% Problem Landing in Board Packets This Week

The headline your board saw

The Conference Board published a report on April 22 that will show up in someone’s board packet within the next two weeks. The headline number: in the S&P 500, the share of companies disclosing AI as a material risk jumped from 12% in 2023 to 83% in 2025.

That number is not the interesting one.

The interesting one sits three paragraphs in. Across the same S&P 500 boards, director-disclosed AI expertise rose from 1.5% in 2021 to 2.7% in 2025. Technology expertise over the same window went from 20% to 51%. Cybersecurity from 15% to 27%. AI barely moved.

So the risk disclosure grew almost sevenfold, and the expertise needed to oversee it grew by a rounding error. That gap is the thing boards are now being measured against, whether they know it or not.

What it actually means

This is the first public, comparable number that puts hard data on something Series C founders have been feeling for six months. The enterprise customer’s procurement team keeps sending AI governance questionnaires. The lead independent director keeps asking what “responsible AI” means, and then nodding at whatever answer gets served up. Legal flagged the KPMG/INSEAD AI Governance Principles for Boards that published April 14 and asked if the board should “look at that.”

The Conference Board report changes the ambient pressure. Before this, a director could reasonably say “we are all figuring this out together.” That defense weakens the minute a regulator, an enterprise buyer, or a plaintiff’s lawyer can cite a specific institutional benchmark for how much expertise should be present in the room.

And the expertise gap is only one number in the report. Only 9% of surveyed executives say their board is very prepared for AI regulation. 26% plan to strengthen AI board education. 21% plan to add a director with AI or technology expertise. In other words, the boards themselves are telling anyone who asks that they are behind.

Grant Thornton’s 2026 AI Impact Survey landed the other half of this picture last month. 78% of business executives lack strong confidence that they could pass an independent AI governance audit within 90 days. The Conference Board data says boards know they lack expertise. The Grant Thornton data says executives know they lack readiness. The same people are telling two different publications the same thing.

That convergence is what makes this a board-packet item, not a headline. A Series C board looking at this question is no longer looking at it in a vacuum. It is looking at it with a benchmark.

Three questions your board will ask

1. “Where does AI oversight sit on this board today, and how do we know that is enough?”

The honest answer is almost never “the full board” or “the audit committee alone.” It is usually nowhere specific, which is why the topic feels ungoverned. The KPMG/INSEAD principles published April 14 give a useful shape: five areas covering strategic oversight, active technology and security oversight, workforce transformation, trustworthy AI standards, and the board’s own work. Pick one committee. Name the owner. Put it on the charter. That is the defensible answer, and it is something a board can decide this quarter without hiring anyone.

2. “Can we pass an AI governance audit in the next 90 days?”

If the Grant Thornton number is right, four out of five peers would say no. The useful move is not to get to yes in one quarter. It is to know exactly which five things would fail an audit, write them down, and show a credible plan to close them. Enterprise buyers and boards both respond to specificity much better than to reassurance. A one-page readiness baseline beats a ten-page governance philosophy in every procurement conversation I have seen.

3. “Are we going to add an AI-literate director, or build expertise on the board we have?”

This is the quiet question. 21% of surveyed boards say they plan to add a director with AI or technology expertise. That market is about to get competitive. If a Series C company does not want to compete for the same five names every peer is calling, a structured three-session AI briefing series with outside experts ships faster, costs less, and the current directors tend to thank you for the insult-free version.

We are not behind the pack. We are with the pack, and the pack is about to get sorted.

The 60-second brief

If there is one minute with the board, here is the minute. The Conference Board released public data on April 22 showing that 83% of S&P 500 companies now disclose AI as a material risk while only 2.7% of their directors claim AI expertise. Grant Thornton says 78% of executives cannot confidently pass an independent AI governance audit today. These numbers are about to sit in every enterprise customer’s procurement questionnaire and every activist investor’s letter. The proposal is simple. Name the committee that owns AI oversight. Commission a one-page readiness baseline in the next 30 days. Decide by the next meeting whether the plan is to build expertise on the current board or recruit for it.

What to watch

Two things are worth tracking in the next 60 days. First, whether institutional investors start citing board AI expertise in their proxy voting guidelines the way they did with cybersecurity between 2020 and 2022. Second, whether the top three enterprise customers add explicit AI governance attestations to procurement packets before their next renewal. One or both will show how fast this compresses. Neither is cause for panic. Both are reason to have a draft answer ready before someone asks for one.