Boards lag on AI oversight as adoption speeds up

New analysis from finance governance researcher Kailash Nath Sadangi says corporate boards are approving AI-informed decisions faster than they can scrutinize them. The gap raises accountability and risk questions as companies increasingly rely on AI in financial and strategic decisions.

Why it matters: - Corporate boards are moving faster on AI-enabled decisions than on the controls needed to oversee them. - The gap raises questions about accountability when AI influences forecasting, credit risk and capital allocation. - Weak oversight can leave boards approving decisions they do not fully understand.

What happened: - Finance governance researcher Kailash Nath Sadangi released analysis warning of a widening disconnect between AI adoption and board oversight. - The analysis focuses on corporate boards approving AI-informed financial and strategic decisions without fully scrutinising how those decisions are generated. - Sadangi is a senior finance executive with Group CFO experience across Australia, the GCC and international markets, and a doctoral researcher focused on CFO-centred governance of AI-enabled decision-making.

The details: - Deloitte's 2025 Global Boardroom Program survey covered nearly 700 directors and executives across 56 countries. - Almost half of respondents said artificial intelligence is not yet formally on their board's agenda. - Separate industry research found that a majority of directors and executives describe their own AI knowledge as limited to non-existent, even as most of their organisations are actively deploying the technology. - PwC's 2025 Annual Corporate Directors Survey found that only around a third of directors said their boards had meaningfully incorporated AI into oversight roles. - In that same survey, two-thirds of directors said boards should be spending more time on AI. - ISS-STOXX reviewed thousands of Russell 3000 and S&P 500 companies and found that only a small fraction disclosed having even one director with specialised AI skills. - Sadangi's analysis draws on the upper echelons framework from Hambrick and Mason, which argues that senior leaders' judgement, experience and discretion materially shape strategic outcomes. - The analysis argues that AI does not remove executive discretion; it makes the accountability chain longer and less visible. - The research uses four questions to test whether a board has real AI oversight: whether directors can explain how a material AI-informed recommendation was generated; whether a named owner exists when human judgement and model output diverge; whether the audit or risk committee tests AI-driven outputs directly; and whether AI oversight is a standing item on the board's governance calendar. - Sadangi said boards lose part of their oversight function when they cannot identify who is responsible if a model's recommendation is wrong. - Sadangi said boards do not need to become technical experts, but they do need clearer decision rights, assurance frameworks that cover model-driven outputs and frank discussion about where AI sits in the decision chain. - Sources cited in the analysis include Deloitte Global, PwC, ISS-STOXX and the Academy of Management Review article on upper echelons theory. - The release included links to Deloitte's boardroom AI report, PwC's corporate directors survey reference, ISS-STOXX's oversight review, and Sadangi's LinkedIn profile.

Between the lines: - The analysis suggests many boards are treating AI as a business tool before treating it as a governance issue. - That creates a risk that oversight lags behind adoption, especially where AI outputs shape material financial decisions. - The problem is less about technical expertise and more about whether boards can trace responsibility through the decision process.

What's next: - Boards that want tighter oversight may need to add AI to formal agendas, test outputs directly and assign clear ownership for AI-related decisions. - Governance frameworks will likely face more pressure to document how humans and models interact in material decisions. - As AI becomes more embedded in corporate operations, scrutiny of board competence on the topic is likely to intensify.