Companies are moving beyond experimentation and embedding AI into workflows, operating models and decision-making. But as adoption spreads, a second challenge is emerging: the governance structures designed to oversee the technology are still catching up.

In Deloitte’s global survey of 695 board members and C-suite executives, 53% said their organizations need to accelerate AI/GenAI adoption, while only about a quarter were satisfied or very satisfied with their current pace.

The pressure to move faster is understandable. PwC finds that AI is increasingly affecting far more than technology strategy, extending into capital allocation, workforce design, operating models and risk. Yet adoption alone does not guarantee returns: 43% of CEOs report either revenue growth or cost reductions from AI initiatives, while 56% say their companies are realizing neither.

For boards, AI oversight is therefore becoming both a risk and value-creation question. Directors need to understand not only whether AI is being deployed safely, but whether management is investing in the right applications and translating adoption into measurable outcomes.

The challenge is that board readiness has not necessarily kept pace.

Deloitte finds that 66% of respondents see limited AI knowledge and experience at the board level, while 31% report that AI is not yet on the board agenda. Another 33% point to insufficient board time dedicated to AI, and 40% say board composition is being reconsidered in response to the technology.

PwC’s research reinforces that gap: 71% of directors identify AI as the board capability most in need of strengthening. But the answer is not necessarily filling boards with technologists. Instead, directors need enough AI fluency to challenge management, understand where the technology is changing the business and oversee the associated strategic and financial risks.

Boards are already trying to build that fluency. 59% are relying on independent AI learning by board members, according to Deloitte, while 48% provide foundational education and 36% regularly engage external specialists. Just 10% are adding AI expertise through new directors.

That distinction matters because AI fluency is not the same as AI governance. Understanding the technology does not automatically answer the questions that emerge when AI begins influencing consequential decisions.

Who owns an AI-informed decision? Which outputs require human validation? How should AI-generated analysis appear in board materials? And when should a human be required to review or override a recommendation?

Those questions become particularly important for private equity, where AI is increasingly capable of influencing diligence, risk assessment, scenario analysis and the information ultimately presented to investment committees and portfolio company boards.

Datasite’s AI Governance Guide for Boards argues that as AI-generated analysis moves into the boardroom, its involvement needs to remain visible. Directors need to understand where AI was used, what assumptions shaped the output, how that output was validated and who ultimately owns the resulting decision.

The objective is not to remove AI from high-impact decisions. It is to ensure those decisions remain explainable, challengeable and attributable to a human decision-maker.

PwC reaches a similar conclusion around the rise of AI agents. As machines take on more execution, human roles are likely to shift toward judgment, exception handling and risk oversight. Boards and management therefore need clear boundaries around when humans must approve, review or override AI-driven actions.

For PE-backed companies, that turns AI governance into something broader than technology policy.

A model might surface an anomaly during diligence, identify patterns across thousands of documents or generate scenarios around a strategic decision. But determining whether a risk should change valuation, whether management can address it or whether the potential return justifies taking it remains a judgment call.

The governance question is moving from “Can we use AI here?” to “What decision rights are we willing to give it?”

The mandate does not stop at managing downside risk.

PwC argues that boards should treat AI as an enterprise transformation rather than a standalone technology initiative. That means overseeing where capital is deployed, which use cases deserve priority and whether AI investments are generating measurable business outcomes.

For private equity boards, that framing should feel familiar. Not every AI initiative deserves funding. Some may materially improve productivity, margins or competitive positioning; others may remain expensive experiments.

The role of the board is ultimately the same as with any major transformation: challenge the strategy, scrutinize the investment case, monitor the outcomes and hold management accountable.

What AI changes is the speed and complexity of the decisions being made.

As adoption accelerates, the strongest boards may not be those with the deepest technical knowledge. They may be those that know where to challenge AI, where to trust it and where human judgment must remain final.

Datasite’s AI Governance Guide for Boards explores how directors can identify AI-generated analysis, challenge assumptions, maintain human accountability and create a defensible record around AI-informed decisions.

[Download the AI Governance Guide for Boards →]