The Larridin State of Enterprise AI 2025 report surveyed more than 1,000 enterprise leaders. The era of AI experimentation is over. The era of accountability has begun.
89% of enterprises have now adopted AI tools. Only 23% can accurately measure their return on investment.
That gap does not exist because AI fails to deliver value. It exists because most organizations deployed AI tools before building the infrastructure to measure, govern, or control them.
The data shows how deep the exposure runs. 67% of enterprises lack complete visibility into which AI tools their employees use. 45% of AI tool adoption happens entirely outside formal IT procurement. The average enterprise runs 23 different AI tools, and most have never been security-reviewed, data-classified, or formally approved.
This shadow AI phenomenon mirrors the early days of cloud adoption. The difference is how many people now use these tools and how they use them, which raises the risk. When employees adopt AI tools on their own, they often share sensitive data, including customer records, financial models, and proprietary documents, with systems that legal, security, and compliance have never evaluated.
Enterprises recognize the need for AI governance. 78% call it a top-three priority. Only 31% have actually built a framework. Many of the frameworks that do exist have to be redefined every year, because the original design never accounted for how fast AI moves.
That disconnect between stated priority and real execution is where most enterprise AI risk lives today. A governance policy that exists as a document is not the same as governance that functions as infrastructure.
Infrastructure-level governance means being able to answer four questions on demand:
Most enterprises cannot answer these with confidence. As the EU AI Act's high-risk obligations enter full enforcement and model risk management expectations extend to AI deployments, this gap moves from a governance concern to a regulatory and financial exposure.
The Larridin report identifies a clear divide between organizations that built AI governance as infrastructure and those that did not. The performance gap is significant.
Organizations with strong measurement and governance frameworks report:
These results come from the minority of enterprises that implemented all three components the report identifies as critical. Usage analytics, meaning who uses what, how often, and for which tasks. Outcome metrics, connecting AI usage to business results. And comparative analysis, measuring the delta between AI-enabled and traditional workflows.
Organizations that implement all three report 5.2x higher confidence in their AI investments and 3.8x higher continued investment rates than those without measurement frameworks.
The research is clear on what works. Establish visibility. Build proficiency through structured training. Implement measurement frameworks that connect AI usage to business outcomes. The organizations doing this are not moving slower. They are moving faster, with less rework, fewer compliance gaps, and measurable results.
The question is no longer whether to invest in AI. It is whether the foundation exists to make that investment accountable.
Teleion AI works with enterprises to build AI governance as infrastructure, not compliance theater. Our work spans the six layers that make governance real:
We have built these frameworks inside some of the most demanding enterprises in the world, including standing up enterprise-wide responsible AI governance for a Fortune 10 technology company and delivering ISO/IEC 42001 certification for a software company ahead of a typical schedule.
If your organization is somewhere in the gap between AI adoption and AI governance, we can help close it.
Source: Larridin State of Enterprise AI 2025.
Start with a focused AI Governance Diagnostic. It is a 2 to 4 week assessment that inventories your current AI tool footprint, surfaces unreviewed data flows, and delivers a prioritized remediation roadmap.
Let's talk!