Who Is Watching the Algorithm? Why AI Adoption Is Outrunning Board Oversight

Organisations are putting artificial intelligence into decision making faster than their governance systems can keep up. That is the central finding of new research from the Chartered Governance Institute UK & Ireland, which combined 16 cross sector roundtables involving 61 participants with survey responses from around 100 governance professionals. The sample spans corporate, public, charity and education settings, and the message is consistent: adoption is moving ahead of the structures meant to oversee it.

AI Is Everywhere, but Boards Often Cannot See It

According to the study, AI is now widely used for tasks ranging from reporting and analysis to compliance and governance processes themselves. Yet many boards lack a clear picture of how it is being used across their organisations, especially where adoption happens informally, outside approved channels. The report found that so called shadow AI occurs at least occasionally in most of the organisations surveyed. In other words, a tool can shape a decision long before anyone responsible has heard of it.

Governance Without Infrastructure

The research suggests many organisations use AI without the frameworks, assurance mechanisms and oversight capability needed to govern it. Formal infrastructure remains thin: few have frameworks to evaluate pilots, to measure how AI is scaled, or to set out clear accountability. Valentina Dotto, policy adviser at the Institute and author of the report, notes broad agreement across sectors that adoption is advancing faster than the ability to govern it, and calls for oversight of where AI is used, how it influences decisions and what controls keep outcomes explainable and accountable.

Awareness Is Not Competence

Boards, the report says, are fairly aware of AI, but practical competence lags behind. Training received by respondents leaned heavily toward basic awareness, and its average rating for preparing people to challenge AI outputs and fold them into governance processes was just 2.4 out of five. Almost three quarters of respondents back making AI literacy a mandatory part of director induction and professional development, a sign that the gap is felt by the people closest to it.

When Systems Start to Act on Their Own

The issue grows more pressing as systems become more autonomous. The report warns that agentic and semi autonomous tools, able to initiate actions with less human input, will need clearer escalation protocols, stronger controls and explicit accountability structures. It also describes a change in what governance means: less about reviewing individual outputs, more about overseeing the systems that generate and influence them, with greater weight on transparency, auditability, assurance and organisational controls. The Institute asks organisations to strengthen visibility over AI use, embed AI within governance and risk frameworks, build board oversight capability and keep accountability clear wherever AI is deployed.

What It Means for Boards and Leaders

Responsibility stays with boards and senior leaders, however automated the process becomes. That makes this a question of people as much as policy: who sits at the table, what they can actually challenge, and how decisions are owned. It is territory close to Euren’s work, from governance and business advisory to the assessment of leaders and the search for the executives and board members able to oversee complex change. Companies that treat AI oversight as a skill to be built, and not a box to be ticked, will be better placed to answer the question the research quietly raises: when an algorithm influences a decision, who is accountable for it?