Request a diagnostic
Twelve weeks. That’s roughly how long the AI Native Board Standard estimates it takes a board to reach entry-level AI fluency, if it puts in the work. Developing the judgment to challenge major AI investments and assess model-related risks can take several more quarters.
Most boards haven’t started on that. NACD’s 2025 Public Company Board Practices and Oversight Survey found 62% had carved out full-board agenda time for AI. Only 27% had put AI oversight into a committee charter, and fewer than 18% had an education plan for directors. Management AI metrics? 6%.
That gap would matter less if companies weren’t telling investors AI is a risk. The Conference Board's April 2026 report counted 83% of the S&P 500 flagging AI as a risk in 2025, up from 12% in 2023, so when something goes wrong (and eventually something will), directors are going to be asked what they did about it.
Here’s a simple test. Imagine a board with two former CTOs on it and no mention of AI in any committee charter. Is it AI-ready?
Not really. It has two AI-ready directors, and when they rotate off, the board’s AI capability walks out the door with them.
That’s why the AI Native Board Standard grades the people and the machinery separately. At the entry level, AI Fluent, directors can follow an AI discussion without someone translating for them. The board itself has handed AI oversight to a named committee in writing, gets AI information on a regular schedule and holds AI discussions where every director speaks up, including the ones who’d happily leave it to the CTOs. A board below that level has yet to demonstrate the collective fluency and oversight practices required by the standard.
The most common mistake? Booking a generic AI course before anyone knows where the gaps are. Directors sit through a morning of slides on large language models and walk out knowing about as much about their own board as when they walked in.
Different frameworks look at different things here. Korn Ferry’s research on AI-savvy boards focuses on the individual: it describes five traits those directors tend to share, including enough technical fluency to ask management sharp questions and a habit of catching data, bias and compliance risks early. Christian & Timbers’ AI Board Readiness Assessment looks at the board as a group. It evaluates collective capability across seven areas, from AI strategy to board oversight, then asks a practical question about each gap: fix it in the boardroom, or fix it in management?
You want both lenses. A single director who carries every AI conversation is great, right up until she misses a meeting.
The AI Board Readiness diagnostic tests people and paperwork together. Each director spends about 20 minutes on a confidential 24-question self-assessment across six areas, and a separate 20-question audit goes through the board’s AI governance documents. Then the full board spends 90 minutes on a disguised AI decision from a comparable company while a facilitator scores the discussion.
You learn a lot from who speaks and who doesn’t. The chair gets a report that ends with a 12-month plan, named owners and a re-assessment date.
Not ready for that yet? The 5-minute self-check or the 10-point readiness checklist will give you a rough picture first.
Under Delaware’s Caremark doctrine, directors can face liability if they make no good-faith effort to put reporting systems in place for mission-critical risks. If AI is mission-critical for your company, and for a lot of companies it now is, the board should be able to show what oversight it put in place. A written committee mandate is a good start. Minutes showing how directors questioned management and followed up on identified risks provide further evidence of that oversight. Caremark and AI get a fuller treatment in a separate article.
Big companies have moved quickly on this. EY’s Center for Board Matters found about 40% of Fortune 100 companies disclosed giving AI oversight to at least one board committee in 2025. The year before, it was 11%.
The audit committee was the most common pick, which makes sense on paper since audit already handles risk. Its agenda is packed, though. EY also found that companies putting AI with a technology committee or the nominating and governance committee tended to write more specific charter language. Whichever committee you choose, the charter should spell out what management reports, how often, and what kind of AI incident goes straight to the full board. The big strategic questions about AI stay with everyone.
Deloitte asked nearly 700 board members and executives across 56 countries about this in its 2025 survey. Two-thirds said their boards had limited or no AI knowledge or experience. That’s progress, believe it or not, since the previous survey put it at 79%.
Training directors one at a time tends to backfire. You end up with one or two fluent people and a room full of polite nodding. It goes better when the whole board learns together, using the company’s own use cases. Korn Ferry’s interviews with directors describe quarterly briefings, simulations, demo sessions and peer exchanges, and some directors go further and use AI tools to draft their own updates or rehearse shareholder questions.
Every director should try some version of that. Reading about hallucination is one thing. Watching a model confidently cite a report that doesn’t exist is another, and directors who’ve seen it happen tend to ask better questions about reliability. The AI Fluent standard builds hands-on use in, along with working knowledge of terms like token, RAG, agent and inference cost, and puts that level at roughly 12 weeks.
[Table 1]
Pilot counts make a board pack look busy without telling directors much. MIT's Project NANDA reported in its 2025 GenAI Divide study that most enterprise generative AI pilots it examined had yet to show measurable P&L impact, which gives directors a good reason to ask management one blunt question: how do you tell an experiment from a deployment that’s paying for itself?
A better board pack starts with a list of AI use cases in production. Each one needs a named owner and the P&L line it affects. Incidents and near misses come next. After that, the board wants to see exposure under rules like the EU AI Act and spend measured against return. Lock the cadence into the committee charter so the reporting shows up every quarter whether or not AI is in the headlines.
Then there’s risk appetite, which boards tend to leave vague. Remember Air Canada? In 2024, a Canadian tribunal ordered the airline to pay up after its website chatbot gave a passenger wrong advice about bereavement fares. The payout was well under C$1,000, but the case showed that a company can be held responsible when its customer-facing chatbot gives information that conflicts with its own policies.
Agree with management, in writing, which high-risk AI uses require additional review before launch, who has approval authority and which incidents must be escalated to the board. For the first full briefing, the 25 questions for management make a decent agenda.
Can’t the board just hire an AI expert? Mostly, no. There aren’t enough of them.
The Conference Board found directors with disclosed AI expertise went from 1.5% of S&P 500 directors in 2021 to just 2.7% in 2025. Only 23% of the 130 executives it surveyed called their own board highly fluent in AI.
Composition still matters, though. MIT CISR studied U.S.-listed companies with more than $1 billion in revenue and found that those with digitally and AI-savvy boards averaged a return on equity 10.9 percentage points above their industry average. That’s a correlation, so it can’t prove one new director will move the numbers. It’s hard to ignore all the same.
A reasonable rule of thumb is to recruit when AI sits at the center of the product or operating model, or when the board is signing off on AI capital commitments big enough that no training course will prepare anyone to evaluate them.
The candidate pool is narrower than people expect. Think former CTOs, Chief AI Officers and Chief Data Officers, plus VPs of AI, Heads of AI and VPs of Engineering who’ve shipped AI products. Most of them will need some help with governance. Seasoned CEOs and CFOs are the mirror image, with plenty of boardroom credibility and thin AI evidence.
One warning for nominating committees. EY found that 44% of Fortune 100 companies cited AI expertise in at least one director biography or identified it as a skill sought for the board in 2025, up from 26% in 2024. That counts companies making the claim. It tells you nothing about depth, which is exactly what the Step 1 diagnostic is for.
Without a published benchmark, “AI-ready” ends up meaning whatever the most confident person in the room says it means. Nobody can defend that in front of investors.
The AI Native Board Standard has three levels. AI Fluent directors understand what management is telling them, and an AI Forward board can go a step further and pressure-test the plan and price the risk. AI Native sits at the top, where AI becomes a standing input to how the board decides.
The indicative timelines are about 12 weeks, two quarters and four quarters. Some boards will take longer, especially if they’re starting from a low base or struggling to document what they’ve changed, and a year of work doesn’t guarantee any particular level. Certification against a documented version of the standard gives boards a consistent basis for measuring their progress over time. Run the diagnostic again at year end and you’ll see whether the work moved anything.
The AI Native Board Standard estimates about 12 weeks to reach AI Fluent. Its indicative periods for AI Forward and AI Native are around two and four quarters, though where a board starts, and how well it documents new practices, will change the pace.
[Table 2]
No. There aren’t enough directors with deep AI experience to go around, and a lone expert often ends up doing the whole board’s thinking on AI. Train the group first. Recruit for whatever gap the diagnostic says training can’t fix.
Usually the audit committee, at least among Fortune 100 companies. Technology committees and nominating and governance committees are the other options, and EY found companies using them tend to write more detailed charter language. Strategy stays with the full board either way.
A list of AI systems in production, with an owner and a P&L line for each. Then ask how management reports incidents and which regulations cover the company’s AI use.
