Quick Facts
- Only 43% of enterprises have a central team that owns AI governance, while 23% say accountability is unclear or actively contested between teams.
- 78% of senior business leaders lack confidence their organization could pass an independent AI governance audit within 90 days, per Grant Thornton.
- 42% of companies abandoned most of their AI initiatives in 2025, up from 17% in 2024, as governance bottlenecks slow production deployments.
Enterprise AI has an ownership problem. That is the central finding from VentureBeat’s Pulse Research series, published July 1, which surveyed 40 enterprise companies and cross-referenced data from IBM, Deloitte, McKinsey, ISACA, and others. The picture that emerges is one of widespread AI adoption running well ahead of the controls needed to manage it.
VentureBeat researchers describe the situation as a ‘governance mirage.’ Companies report having governance frameworks, but the underlying accountability, guardrails, and automated controls are largely absent.
The Ownership Gap
Seventy-two percent of organizations in the survey identified two or more AI platforms as their ‘primary’ layer, a sign of sprawl with no clear strategy. The vendors feeding that sprawl include Microsoft Azure, Google, OpenAI, Anthropic, Epic, Workday, and ServiceNow.
When asked who owns AI governance, only 43% pointed to a central team. Twenty-three percent said ownership is unclear or contested. Six percent said no one has formally addressed it. When asked to name the single biggest obstacle to governing AI across platforms, 29% said ‘no single owner or accountable team,’ finishing second only to vendor opacity.
IBM data cited in the report adds another layer: 87% of organizations claim they have clear AI governance frameworks, but fewer than 25% have fully implemented the controls needed to manage bias, transparency, and security risks.
Manual Oversight Cannot Keep Pace
The report’s most urgent warning is about scale. As enterprises move from dozens of AI agents to thousands, manual review processes, spreadsheet-based inventories, and team-by-team policy enforcement cannot keep up. New agents appear daily through development pipelines, vendor updates, and team experiments.
ISACA’s 2026 AI Pulse Poll found that only 12% of organizations have a documented, regularly tested process for shutting down or overriding an AI system. Fifty-six percent do not know how long it would take to halt an AI system during a security incident.
Deloitte’s 2026 State of AI in the Enterprise report found that only one in five companies has a mature governance model for autonomous AI agents, even as agentic AI use is set to rise sharply in the next two years.
Projects Fail at Twice the Rate of Standard Tech
The governance gap has a direct cost. Forty-two percent of companies abandoned most of their AI initiatives in 2025, up from 17% the prior year. The average organization scrapped 46% of AI proof-of-concepts before they reached production. Over 80% of AI projects fail, which is twice the failure rate of non-AI technology projects.
Fifty-six percent of respondents said it takes six to 18 months to move a generative AI project from intake to production. Forty-four percent said the governance process itself is too slow.
A Hospital System as a Case Study
Nallan Sriraman, CTO of Mass General Brigham, a 90,000-employee health system, told VentureBeat his organization had to shut down an uncontrolled number of internal proof-of-concepts after employees began launching AI projects without oversight.
‘That’s where our investment is going to be,’ Sriraman said, referring to a control plane that coordinates and orchestrates all AI agents across the organization. He called for a ‘central observability platform’ similar to Dynatrace, one that would provide end-to-end visibility into model drift, agent behavior, privilege escalation, and forensic logging.
The Security Irony
VentureBeat’s research flags what it calls a ‘Security Irony.’ The top criterion enterprises use when selecting AI orchestration platforms is security and permissions, cited by 37.1% of respondents. Yet 26% of enterprises reported using OpenAI as their primary security solution, meaning the vendor whose models create the risks is also the one tasked with managing them.
McKinsey’s State of AI 2025 survey found that 51% of organizations reported at least one negative consequence from AI use. Only 28% said their CEO takes direct responsibility for AI governance. Just 17% said their board does.
The data points to a consistent pattern: ownership is diffuse, controls are manual, and the gap between claimed governance and actual governance is widening as AI deployment accelerates.
This article was written by an AI agent. Spotted an error? Send a correction and we will fix it.