Here is a finding from Deloitte's 2026 State of AI in the Enterprise report that should stop every governance leader in their tracks: organisations where senior leadership actively shapes AI governance achieve significantly greater business value than those delegating the work to technical teams alone.
Not marginally better. Significantly better.
If that is true, and 3,235 surveyed global leaders suggest it clearly is, then the AI governance conversation is taking place in the wrong room. Most frameworks are designed, built and owned by technology and compliance teams. The people with the greatest influence over whether governance actually works are rarely meaningfully involved.
McKinsey's 2025 State of AI survey found that 88 per cent of organisations use AI in at least one function. Only 39 per cent demonstrate any EBIT impact. Only 21 per cent have fundamentally redesigned workflows as a result. The research consensus across McKinsey, Deloitte, PwC and MIT is consistent: the primary barrier to AI value is not technical. It is leadership, culture and change management. Governance documents do not govern. People do.
Why Is AI Governance Failing?
The honest answer is that most governance frameworks are not designed to be adopted. They are designed to be completed.
Consider the typical implementation. A policy document is drafted, reviewed by legal and compliance, approved by the governance committee and published on the intranet. Training modules are assigned. A governance register is opened. Three months later, the same business units using AI tools without oversight are still doing so, now with the convenience of being able to point to a framework that technically addresses the situation.
McKinsey identifies the redesign of workflows as the single biggest predictor of EBIT impact from AI. Yet only 21 per cent of AI-using organisations have done this. Workflow redesign is a leadership act. It requires executives to make decisions about how work is done, what AI systems are accountable for and how performance is measured in an AI-augmented environment. No governance document makes those decisions. Leaders do.
The data on leadership accountability is stark. Only 28 per cent of organisations say their CEO takes direct responsibility for AI governance oversight (McKinsey). Only 17 per cent report that their board oversees it directly. At most organisations, AI governance is delegated to the CIO, to a Chief Data and AI Officer or to a newly formed AI Centre of Excellence. The intent is good. The outcome is a governance function that lacks the authority to drive the changes that would actually make it work.
Three Leadership Behaviours That Predict AI Governance Success
Setting explicit accountability for AI outcomes. The organisations where AI governance works have leaders who treat AI systems as business assets with owners, not IT assets managed by technology teams. When a business unit leader is accountable for the outcomes of the AI systems their team uses, the governance conversation changes from "are we compliant" to "is this working and who is responsible when it does not?"
Modelling AI literacy. McKinsey's list of AI scaling best practices includes senior leaders who are actively engaged in driving adoption, including role-modelling the use of AI. This does not mean executives becoming data scientists. It means leaders who can engage credibly with the governance questions AI raises: what decisions is this system making, who has reviewed it, how are we monitoring it and what is the escalation path when it behaves unexpectedly? Leaders who cannot engage with these questions cannot hold their teams accountable for answering them.
Making governance visible in regular communication. Deloitte identifies regular internal communication about AI value creation as a key scaling practice. The same principle applies to governance: when leaders talk about AI governance, not as a compliance obligation but as how the organisation approaches AI responsibly, it becomes part of the culture rather than a parallel function that operates separately from the business.
What Governance as Culture Looks Like in Practice
The organisations where AI governance culture has taken hold share observable characteristics.
AI governance accountability appears in performance frameworks, not as a box-ticking criterion but as a genuine measure: did this leader ensure their team's AI deployments went through the appropriate review process? Did they escalate risk concerns appropriately? Did they act when monitoring flagged anomalous outputs?
AI literacy is treated as a leadership development priority, not a technical training requirement. The emphasis is not on understanding model architecture. It is on understanding governance responsibilities, risk indicators and the questions a leader needs to ask of any AI system their team is operating.
AI governance reporting is regular and substantive at board level. A Dataiku survey of 600 CIOs found that nearly all now brief the board on AI performance at least quarterly and nearly half do so monthly. When AI governance performance is reviewed on that cadence, it enters the same accountability framework as financial performance and operational risk. It becomes real. Gartner's research found that 45 per cent of organisations with high AI maturity keep their AI initiatives live for at least three years, compared to only 20 per cent among lower-maturity peers. Governance sustainability is a leadership outcome.
The Change Management Steps Most Organisations Skip
Engaging the people who will be governed, not just the people who will govern. Governance frameworks designed by governance teams for governance teams create processes that governance teams understand and everyone else works around. Business units that will submit AI use cases, conduct risk assessments and operate AI systems in production need to be involved in designing the processes they will use. Participatory design is the difference between a framework that is used and one that is not.
Building a communication cadence, not just a training programme. A one-off training module does not change culture. Regular communication from senior leaders about how AI governance is working, covering what has been approved and why, what has been declined and why and what the organisation is learning, signals that governance is a genuine business priority, not a compliance obligation to satisfy and forget.
Measuring adoption, not just completion. Most governance implementations measure output: how many policies are published, how many training modules are completed, how many use cases are registered. Few measure meaningful adoption: are the right people using the governance process at the right time in the AI development lifecycle, and are governance decisions having the intended effect on risk management and value creation?
Measuring AI Governance as a Leadership Outcome
Governance as culture requires governance to be measurable as a leadership outcome, not just a compliance output. Track the proportion of AI initiatives going through the governance process before deployment, not retrospectively after an incident. Track the time from use-case submission to governance decision: is governance a meaningful checkpoint or an obstacle that teams time out waiting for? Track the outcomes of declined use cases: are they being redesigned and resubmitted, or going ahead in shadow mode?
AI governance leadership in Australia faces the additional dimension of a regulatory environment that is maturing rapidly. Organisations that build the leadership culture now, before APRA, ASIC or sector-specific AI regulation requires it, will be significantly better positioned than those waiting for external mandates to force the conversation.
What This Means for Your Organisation
At Trusenta, the engagements where we see the fastest and most durable governance adoption are consistently those where a senior leader, a CEO, CIO or Chief Risk Officer, has made AI governance a personal leadership priority, not just a project they have sponsored. That leadership signal changes everything downstream: how business units engage with governance, how seriously governance decisions are taken and how quickly the culture shifts from seeing governance as overhead to seeing it as how the organisation operates responsibly at scale.
The framework matters. But the framework without the leadership culture around it is a document on an intranet. The goal is governance as operational practice: embedded, owned and modelled from the top.
Key Takeaways
- Deloitte's 2026 research confirms that senior leadership involvement in AI governance is a significant predictor of business value from AI
- Only 28% of organisations have their CEO directly responsible for AI governance oversight, a leadership gap and not just a governance one
- Three leadership behaviours predict governance success: explicit accountability for AI outcomes, AI literacy and visible governance communication
- The biggest change management failures are excluding business units from governance design, relying on one-off training and measuring completion rather than adoption
- Governance as culture means it appears in performance frameworks, leadership development and regular board reporting, not just policy documentation
- The organisations generating real AI value treat governance as a leadership operating model, not a compliance exercise
How Trusenta Can Help
AI Governance Services: Trusenta's AI Governance services are designed to move organisations from policy documentation to operational governance, with frameworks built for adoption by leadership and business units, not just compliance teams.
Fractional AI Officer: For organisations that need a senior AI leader to drive governance as a cultural priority, this service embeds an accountable executive who owns the governance agenda, represents it at board and committee level and models the leadership behaviours that the research identifies as the real differentiators of governance success.
AI Governance Foundations: This 10-day engagement is designed not just to produce governance documents but to establish the roles, decision rights and accountability structures that make governance adoption possible, the operating conditions that allow culture to form around governance rather than around working past it.
Conclusion
AI governance frameworks will not save organisations that treat governance as a compliance function. The research is clear: governance succeeds or fails at the leadership and cultural level, not the policy level. The organisations moving to production, generating measurable returns and managing AI risk effectively are those where senior leaders own the governance agenda, model the behaviours and hold their teams accountable for outcomes, not just for having a framework in place. The question is not whether your organisation has a governance policy. It is whether your organisation has a governance culture.
.png)
.jpg)
.jpg)
