The 3 AI Governance Mistakes Boards Still Make (And How to Fix Them Before It’s Too Late)

Author(s): Shane Culbertson

Originally published on Towards AI.

Artificial Intelligence (AI) is everywhere, yet far too many boardrooms still treat it as a side project. That needs to change. Fast.

As someone who works directly with directors and C-suites, I’ve sat through AI updates that sounded more like product launches than strategy sessions. I’ve also seen boards that never discuss AI at all, even though it’s embedded in their hiring systems, marketing platforms, and credit approval engines. That silence is a mistake, and it’s putting companies at risk.

The good news is that these gaps are fixable. But first, let’s call out three of the most common board-level missteps when it comes to AI oversight.

Mistake 1: AI Still Isn’t on the Agenda

According to Deloitte’s 2025 Global Boardroom Survey, only 17% of boards discuss AI at every meeting. About one-third don’t have it on the agenda at all. In a time when generative AI tools can upend entire customer workflows in a matter of weeks, this is an oversight boards can’t afford.

One-third of boards still don’t talk about AI. That silence isn’t neutral — it’s negligence.

Keeping AI off the agenda means no accountability, no progress tracking, and no structured oversight of risk. AI governance becomes an afterthought, rather than a strategic pillar.

How to fix it:

· Require management to provide a short AI status update at every board meeting.

· Ask how each AI use case aligns with strategic goals.

· Ensure AI is included in your risk register, compliance reports, and internal audit plans.

If AI touches your customer interactions, operations, and compliance risk, then it belongs in the boardroom.

Mistake 2: The Board’s AI Knowledge Gap is Too Wide

More than half of the boards report they have little to no AI fluency. This creates one of two scenarios: either directors are deferring to management mindlessly, or they’re skipping the topic altogether. Both are dangerous.

Without a working knowledge of how AI systems function, how they learn, where they can go wrong, and what their limits are, boards can’t provide meaningful oversight.

How to fix it:

• Schedule foundational AI training at least once a year.
• Bring outside experts to run scenario reviews and Q&A sessions.
• Appoint or recruit at least one director with operational AI experience.

You don’t need to turn your board into a team of data scientists. However, you do need to know what to ask, how to challenge vague answers, and where ethical red flags might be hidden.

Mistake 3: Ethics and Fairness Are Left to the Developers

Too many boards assume that engineers and product managers are handling fairness, privacy, and bias testing. That assumption is flawed. Developers are rarely incentivized or resourced to handle ethics oversight. If the board doesn’t ask about it, it likely won’t get done.

Developers can build AI, but they shouldn’t be expected to govern it alone.

Harvard Business Review identifies a persistent “AI trust gap” tied to black-box decision-making, algorithmic bias, and weak transparency frameworks. That trust gap is already slowing adoption in industries like healthcare and financial services.

Trust isn’t a byproduct of performance. It must be built into the system from the outset.

How to fix it:

• Mandate that every significant AI initiative undergoes an ethical risk review.
• Include bias monitoring metrics and explainability thresholds in project KPIs.
• Build scenarios for when AI fails: Who is accountable? How will decisions be audited or reversed?

A strong AI ethics framework doesn’t just prevent harm, it builds trust with regulators, customers, and the public.

The Bottom Line: Proactive Boards Will Win

Boards that fail to address these mistakes are taking a risk with their reputational standing, compliance exposure, and potential innovation opportunities. Those who act now can shape how AI is applied across the organization, long before regulations or crises force the issue.

AI oversight isn’t optional anymore. It’s part of your job.

Board takeaways:

• AI oversight is no longer optional.
• You don’t have to become a technical expert, but you do need to ask better questions.
• Ethics and fairness must be built into the lifecycle, not bolted on afterward.

This is the moment to lead with intention. Companies that get AI governance rights won’t just avoid trouble; they will also reap the benefits. They’ll build systems people can trust.

Sources & Further Reading

Deloitte Global Boardroom Program Survey — “Governance of AI: A Critical Imperative for Today’s Boards,” Harvard Law School Forum (May 27, 2025). https://corpgov.law.harvard.edu

Harvard Business Review — Bhaskar Chakravorti, “AI’s Trust Problem: Twelve persistent risks of AI that are driving skepticism,” (May 3, 2024). https://hbr.org/2024/05/ais-trust-problem

Mamia Agbese et al., “Governance in Ethical and Trustworthy AI Systems,” e-Informatica (2023). https://doi.org/10.37190/e-Inf230101

Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor.

Published via Towards AI