Boards used to slow AI down. Now they push it faster than most executives can manage. Pano Anthos, founder and managing partner at XRC Ventures, has watched more than 140 portfolio companies face this shift. Glenn Gow, The Scaling Executive Coach, sees CEO clients hit the same handoff of risk. The board hands you a mandate, and you own what happens next.
Quick Answer
Boards now expect CEOs to run point on AI governance, even when the board itself can’t evaluate the systems it’s approving. Glenn Gow tells his CEO clients the fiduciary duty doesn’t disappear just because the board lacks the technical depth to weigh in. It shifts to you. Bring the board hard numbers on risk, spend, and oversight before they ask, and you turn a gap in their expertise into your case for staying central to every governance decision. I’ve coached executives through this exact handoff of risk, and the pattern holds: boards that can’t adjudicate AI vendors still expect someone in the room who can.
The Board Flipped From Skeptic to Demander
Pano has watched boardroom pressure on AI reverse direction entirely. Two years ago, boards asked hard questions before approving new technology. Now they push executives to move before the technology is proven. Pano put it this way: “Now the boards are saying, we need AI in our applications, in our world. So now executives have carte blanche to go run hard and test AI solutions up and down the food chain.”
I tell CEOs to treat that kind of mandate as a blank check with their name already signed on the risk line.
Elias Stahl, CEO and co-founder of HILOS, described the same pressure from a different angle: competitive anxiety instead of enthusiasm. His board isn’t excited about AI. It’s afraid of falling behind. In his words: “The board wants us to use AI. We have to use AI as fast as possible. We’ll fall behind. They’re under a huge pressure to use integrated AI as quickly as possible to show ROI from it.”
I ask every CEO in Elias’s position the same question: what’s the cost of moving eight months faster if the board never asks how you validated it? Most can’t answer yet.
In my work with CEO clients, I now push them to walk into every board meeting with an AI risk summary already prepared. Waiting for the board to raise the topic means you’re already behind. The CEOs who get ahead of this treat the board’s request as a starting point, not a mandate to sprint blind.
Why Boards Can’t Evaluate What They’re Approving
Board oversight only works if the people doing the overseeing understand what they’re looking at. Pano told me most boards don’t. Vendors pitch AI tools faster than any board can vet them, and the board has no reliable way to separate a real solution from a repackaged one. Pano, whose fund at XRC Ventures often gets pulled in to advise boards on exactly this, explained it directly: “No enterprise can possibly absorb all the innovation that’s hitting them… they can’t adjudicate A from B, and so bringing us in as either an LP or a GP to them, or just as an advisor, we can actually help filter that list very quickly.”
Steven Collens, CEO of MATTER, sees the identical gap play out in hospital boardrooms, where the stakes are higher because the tools touch patients directly. Hospital boards, he said, hold back on approving patient-facing AI not because they’ve done the analysis and found it wanting, but because they don’t know how to run the analysis at all: “They’re very reticent to adopt patient-facing AI tools because it’s scary and they don’t really know how to evaluate how these things are built, how robust they are, and what level of risk they’d be taking on.”
Steven’s boards freeze. Pano’s boards push through anyway. Either response leaves the CEO holding a decision the board isn’t equipped to make with them.
I tell clients that boards drowning in vendor pitches will lean on whoever gives them the clearest filter. If that’s not the CEO, it will be an outside advisor, the way Pano’s fund now functions for the boards he sits on. Either way, the filtering happens. The only open question is who does it.
What the Board’s Fiduciary Duty Actually Covers
Bernardo Hernández, co-founder and co-CEO of Pensero AI, defines the board’s job in plain terms, and it has nothing to do with technical fluency. A board exists to protect resources and identify risk. Bernardo said: “Board is the most senior governance body within an organization. At the end of the day, they hold the responsibility of an effective use of the resources, identifying the risks, and just making sure the company is creating the largest possible shareholder value.”
In my coaching work, I’ve seen this fiduciary obligation get misread as a technical requirement. The board needs to know whether the company is spending money well and managing risk well, the same standard it already applies to a factory line or a sales team. Very few boards can answer that question yet about an AI system, which is why I push the CEO to do the translating for them.
I push every CEO I coach to frame AI spending in that same language boards already trust: dollars at risk, dollars protected, and who signed off. Boards don’t need a technical briefing. They need the fiduciary framing Bernardo describes, applied to AI instead of a factory line.
Bernardo also pointed to a second gap that sits right next to the first one: engineering is one of the largest line items on most balance sheets, and boards routinely approve it with almost no way to check whether the spend is producing results. That blind spot on ordinary tech investment is the same blind spot showing up now on AI, just with higher stakes attached.
Legal Precedent Is Already Shifting the Risk
Governance stopped being theoretical the moment courts started ruling on AI-generated decisions. Chris Crowe, founder and CEO of CMBYND, points to a case that should worry every CEO who assumes a chatbot’s mistake is the vendor’s problem. He said: “There’s an Air Canada case recently where a bot had created a product for a client and went all the way up to the court and the court had said, you have to honor that. You have to recognize that, which means accountability now is something that’s sitting on all the organizations.”
Chris also raised a harder question most boards aren’t asking yet: what happens once AI stops answering questions and starts taking actions on the company’s behalf. He said: “How do you manage when you go into an agentic world, manage the oversight of how that’s actually potentially shifting your organization and putting more risk on your organization? You’re accountable for what’s being produced and what’s being communicated.”
I tell every CEO I coach that the Air Canada ruling applies well beyond travel bookings. Any company running an AI system that acts on a customer’s behalf now carries the same exposure. Build the audit trail before a court asks you to produce one.
AI Governance Guardrails Must Come Before Experimentation
Colby Durnin, CEO of CREDE, treats governance as the first step, not a cleanup step. Before any team touches an AI tool, he wants contracts, IP protections, and data boundaries locked down. He said: “Before anybody can use AI, you really need to put stop gaps around your governance. It’s really making sure that your intellectual property is secure, that you’re not violating any of your covenants of your contracts for non-disclosure IP with your clients information.”
Colby’s approach flips the usual order. Most companies let teams experiment first and build governance once something breaks. He puts the AI governance guardrails in place first and lets experimentation happen inside them.
I ask every CEO client one question before they approve any new AI tool: who owns the IP risk if this goes wrong? If nobody in the room can answer in under a minute, the tool doesn’t get approved yet.
Most CEOs I coach resist this because it feels like it slows the company down at exactly the moment competitors are moving fast. In practice it moves the slow part earlier, before money and reputation are on the line.
Board-Level AI Talk Is Not the Same as Team-Level AI Use
Ali Wing, CEO of Oobli, separates the board conversation from every other AI conversation happening inside a company. She said: “How are we talking about it at boards? Because that’s a whole different thing.”
Ali pushed further on what the board conversation should actually cover. It isn’t a status update on which tools employees have adopted. It’s a decision about what the company automates and what it keeps out of public systems entirely. In her words: “Every part of a board’s job is primarily to analyze and assess what’s the primary automation capability of AI, and how do we create environments where we can have things that otherwise can’t be put into public domains to leverage automation.”
I see CEOs make this mistake constantly. They report tool adoption numbers to the board and call it governance. Ali’s distinction is the one I bring into every coaching session on this topic: a board update on team usage and a board decision on data exposure are not the same meeting, and treating them as one leaves the real risk undiscussed.
Reactive Boards vs. Governance-Ready Boards
I’ve watched this split play out with almost every CEO client I coach. Boards that govern AI well share risk with the CEO before anything breaks. Boards that don’t leave the CEO holding it alone.
| Reactive Board | Governance-Ready Board | |
| How AI gets approved | Approved under pressure to show ROI fast | Approved after risk and spend are reviewed |
| Who owns the risk | CEO absorbs it after the fact | CEO and board share it upfront |
| Response to a legal or data incident | Board asks why nobody flagged it | Board already has a documented answer on file |
| What it produces for the CEO | Exposure with no cover | A documented case for staying in every governance decision |
I built this comparison from what I hear in board meetings every week. The companies on the right side of that table aren’t smarter. They just started the governance conversation before the board forced it. The CEOs I coach who move to the right side of that table stop treating board updates as a formality and start treating them as the record that protects them later.
Frequently Asked Questions
Who is responsible for AI governance in a company?
The CEO is responsible for AI governance in practice, even when the board holds formal oversight on paper. Glenn Gow tells clients that boards set expectations, but the CEO builds and runs the governance system that actually protects the company. Bernardo points out boards approve some of the largest line items on the balance sheet, like engineering, with almost no way to check whether the spend is working. AI governance sits inside that same blind spot, which is exactly why the CEO has to own it.
What does a board expect from a CEO on AI adoption right now?
Boards expect speed and proof at the same time, which is a harder standard than it sounds. Elias described his own board pushing for fast integration purely to show ROI, not because they had validated the approach. Glenn Gow coaches CEOs to bring the board a risk and return summary before the board asks for one, so the expectation gets met on the CEO’s terms instead of the board’s timeline.
What is a CEO’s fiduciary duty when it comes to AI?
A CEO’s fiduciary duty on AI is the same duty that applies to any major capital decision: protect resources and manage risk. Ali frames the board’s real job as deciding what the company automates and what data it protects from public systems, not just monitoring which tools teams have adopted. Glenn Gow pushes CEOs to bring that same protect-and-decide framing to every AI budget line, not just the ones the board happens to ask about.
How should a CEO present AI risk to the board?
Present AI risk the way you would present any other capital risk: dollars exposed, dollars protected, and who owns the decision. Pano has seen boards hand out open mandates on AI simply because they can’t evaluate the technical details themselves. Glenn Gow tells CEOs that a clear risk summary is what earns you the board’s trust to keep making these calls.
Can a company be held legally accountable for AI-generated decisions?
Yes. Chris points to the Air Canada case, where a court ruled the company had to honor a commitment its own chatbot made to a customer. Chris also raises the harder version of this question: once AI systems start acting on a company’s behalf instead of just answering questions, the organization carries the oversight risk for what those systems produce and communicate. Glenn Gow tells every CEO client to assume both standards apply to their company today, not someday.
Is the CEO personally liable if the company’s AI system causes harm?
Personal liability depends on jurisdiction and the specific facts, and that’s a question for the company’s lawyers, not a coach. What Glenn Gow tells CEOs is that the company’s liability, which Chris’s Air Canada example makes clear is already real, lands on the CEO’s desk regardless of how the personal liability question shakes out. Treat company accountability as the standard to govern against, not the exception.
What is the difference between board-level AI oversight and team-level AI use?
Board-level oversight covers data exposure, automation strategy, and risk. Team-level use covers which tools employees adopt day to day. Ali draws that exact line. Glenn Gow sees CEOs make this mistake constantly: they report tool adoption numbers to the board and call it governance. It isn’t. The board needs the data and automation conversation, not a usage report.
How does a board evaluate AI vendors when it doesn’t understand the technology?
Most boards can’t evaluate AI vendors directly, which is exactly the problem. Steven has watched hospital boards freeze on approving patient-facing AI tools for this same reason: they don’t know how to evaluate how the systems are built or what risk they carry. Glenn Gow recommends CEOs build a vendor-filtering function inside the company instead of leaving that gap to freeze the board or get filled by an outside advisor, since it keeps the decision and the accountability in the same place.
What questions should a board ask before approving an AI project?
A board should ask three things before approving any AI project: who owns the IP risk, what data the system touches, and what happens if it fails publicly. Colby requires the IP and data answers locked down before any team touches a tool. Glenn Gow adds the third question in every board conversation he coaches CEOs through, since it’s the one boards skip most often.
Should a board require a formal AI governance framework before approving AI projects?
Yes, and the CEOs who avoid painful surprises are the ones who build the framework first. Colby requires contract, IP, and data protections locked down before any team touches an AI tool. Bernardo separately pointed out that boards approve massive engineering spend with almost no visibility into whether it delivers results, which is the same blind spot that hits AI budgets even harder. A framework closes both gaps at once.
Is the CEO or the board ultimately accountable for AI outcomes?
Both are accountable, but unevenly. The board carries the fiduciary responsibility on paper, while the CEO carries the operational and legal exposure in practice, especially now that courts are ruling companies accountable for what their AI systems do or say. Glenn Gow tells CEOs not to wait for the board to define that split. Define it yourself, document it, and bring it to the board before an incident forces the conversation.
What happens if a board pushes AI adoption faster than the company can govern it?
The company ends up carrying risk nobody signed off on, which is the exact position Elias described his own board putting him in. Glenn Gow coaches CEOs to slow the pace without saying no outright: agree to move fast on low-risk applications while insisting on governance guardrails, like the ones Colby puts in place, for anything touching customer data or company IP.
What You Do Next
If your board is pushing AI faster than you can govern it, don’t wait for the next board meeting to raise it. Glenn Gow has coached CEOs through exactly this handoff of risk, and the fastest way to protect yourself is to bring the numbers before the board asks. Schedule a call with Glenn to build the governance case that keeps you in the room where these decisions get made.
