NORBERT BIEDRZYCKI

AI Governance as a Board Level Responsibility: Is the Board Ready?

AI Governance as a Board Level Responsibility: Is the Board Ready?

McKinsey’s State of AI survey published in March last year found that in 28 percent of organizations using AI, the CEO is responsible for overseeing AI governance, and only 17 percent reported that their board was. The same research found that CEO oversight of AI governance was among the factors most closely correlated with higher reported bottom line impact from generative AI, particularly in larger companies.

Those two numbers capture the situation many boards are in. AI is increasingly shaping strategy, operations, risk and workforce. Governance of it mostly sits elsewhere.

Why AI belongs on the board agenda

The consequences of AI decisions now reach every area a board is responsible for.

Strategy. AI is changing competitive positions and business models, as I argued in my series on transformation.

Capital allocation. AI investments are among the largest discretionary spending decisions many companies make, and the evidence of returns is uneven.

Risk. Regulatory exposure under the AI Act and similar laws, litigation, cyber risk, operational failures of autonomous systems.

Workforce. Decisions about automation, reskilling and organizational design.

Reputation. As the incidents of the last two years show, one visible AI failure can damage trust built over years.

In corporate law, directors have a duty to oversee material risks. In the United States, the Delaware courts have made clear in cases such as Marchand v. Barnhill that boards can be liable when they fail to establish any reporting system for a mission critical risk. For a growing number of companies, AI is becoming exactly that.

What competence actually means

The usual response is to call for AI experts on boards. That is useful but insufficient. One expert director can become the person everyone else defers to, which is the opposite of governance.

What boards need is collective competence sufficient to challenge management. In practice that means being able to ask good questions.

Where does AI make or shape material decisions in this company, and who owns each of them?

What value has AI actually delivered, measured in P&L terms, and how do we know?

What are our three most significant AI risks, and how would we detect a failure?

Which suppliers and models are we dependent on, and what happens if one of them fails or changes terms?

How are we preparing our people, and what are we deliberately not automating?

The structure question

Some boards are assigning AI to the audit or risk committee. Others to technology committees. A few are creating dedicated committees. The structure matters less than two things: that AI is discussed regularly, not once a year in a strategy session, and that the board receives information in a form that allows it to challenge, not only to be informed.

From my own experience in supervisory roles, the most valuable board discussion of AI is rarely about the technology. It is about decisions: which ones the company is delegating to machines, on what basis and with what safeguards.

Leadership question: Can a board adequately govern a company whose most consequential decisions are increasingly influenced by technologies it does not fully understand?

Read next

  1. Responsible AI and Corporate Reputation: Can One Algorithm Destroy Years of Trust?

    Ethical AI series, written in 2026

  2. Responsible AI as a Competitive Advantage: Can Trust Become an Economic Asset?

    Ethical AI series, written in 2026

  3. Ethical AI: From Principles to Accountability, 2020–2026