The AI Governance Paradox: Steering AI Without Hindering Innovation

Across Australian enterprises, AI agents are multiplying.

What began as targeted pilots is becoming systemic. Copilots draft reports, autonomous agents resolve IT tickets, and AI systems analyze contracts, trigger approvals and move work across departments. Dozens of AI-enabled workflows are already live in many enterprises. Soon, there will be hundreds more.

The scale is global. Bloomberg Intelligence found 97% of global enterprises reported active AI implementation in 2025, with two-thirds moving beyond evaluating AI models to developing and scaling AI tools.



Yet only 33% of Australian leaders have a clear AI vision. This is the governance paradox: while enterprise AI adoption is surging, confidence and control are lagging.

ServiceNow’s 2025 Enterprise AI Maturity Index, based on 4,473 senior leaders globally, including 560 in Australia, highlights the strain. AI risk and governance are seen as make-or-break factors for AI success. In Australia, data security is seen as the largest barrier to realizing AI’s value.

When governance is treated as an afterthought, AI amplifies existing weaknesses. Leaders struggle to answer crucial questions, like where agents are deployed, what data they have access to, and how decisions can be audited and traced.

ServiceNow’s research shows that only 44% of Australian organizations have made meaningful progress in connecting data and operational silos. Fewer than half report significant advances in formalizing governance and compliance controls.

“As AI scales, so does exposure. Bolting on AI governance after deployment is often too late,” notes Pete Andrew, Group Vice President of ServiceNow Australia and New Zealand. “Simply adding AI to existing silos doesn’t help. It just automates dysfunction faster.”

Forward-looking enterprises are taking a different approach. Orica, the global mining and infrastructure solutions provider, uses ServiceNow solutions to centralize its global IT governance process. By managing AI demand, tracking projects and maintaining visibility into approved tools within a unified system, Orica has ensured that its AI expansion is structured rather than fragmented.

“One of the key things that really made a difference for us was establishing strong governance and securing executive alignment from the outset,” notes Rachael Sandel, Group Chief Information Officer at Orica. “This top-down support and the transparency we’re bringing brings clarity and confidence to our teams to move fast while staying safe, responsible and focused on value.”

The next competitive advantage in enterprise AI will not belong to those who deploy the most agents, but those who design governance and guardrails into their systems from the outset.

That requires integrating data, workflow and security within a single operational backbone: an AI control tower that provides real-time visibility across agents, automated monitoring and enforcement, and built-in auditability. With this control tower, AI deployment and governance can move at the same speed.

“It's not enough for AI to simply work; it needs to be trustworthy,” says Bryan McGowan, Global Trusted AI Leader and US Trusted AI Leader at KPMG. “By activating KPMG AI Trust with alliance partners like ServiceNow, we are taking a significant step forward in AI governance to set a new standard for AI risk management across all industries.” 

In the age of rapid adoption, governance is not a brake on innovation, but the foundation that puts AI to work for Australians, safely and sustainably and at scale.