New Survey from Collibra by The Harris Poll Finds 72% of Tech Decision-Makers Feel AI Initiatives Today Are Falling Short

Nearly Nine in Ten Respondents [87%] Burn Hours Re-Verifying Context for AI Agents, with 76% Facing Roadblocks Moving from Pilot Programs into Production

NEW YORK and BRUSSELS, Sept. 16, 2026 /PRNewswire/ -- Collibra, the enterprise AI control plane, today announced the findings of a Harris Poll survey of more than 300 U.S. adults ages 21+ who are employed full time as data management, privacy, and/or AI decision-makers. The survey - entitled The 2026 Hallucination Tax Report - found that while nine in ten organizations are actively deploying autonomous AI agents, more than three-quarters (76%) have hit critical roadblocks attempting to move from pilot into full production over the past 12 months.

The research also found that 72% of decision-makers agree that when their organization's AI initiatives fall short, the root cause almost always traces back to an unaligned or poor data foundation. These findings also reflect broader industry trends across enterprise technology, where Gartner research reports that at least 50% of enterprise generative AI projects are abandoned after proof of concept due to poor data quality, escalating costs, and inadequate risk controls.

"The AI conversation across the C-suite has fundamentally shifted from the theoretical to measurable business impact," said Felix Van de Maele, Co-founder and CEO of Collibra. "We have found that every enterprise scaling AI today is paying a hallucination tax — a hidden cost of manual oversight, rework, and risk that grows with every new agent put in production. If human workers have to spend countless hours manually validating every response, the business value of automation ultimately disappears, and companies carry the burden."

Notably, the report highlights that the absence of structured context and runtime governance is forcing organizations back into costly manual supervision, and, in fact, 87% of decision-makers say their teams regularly re-verify that the context their agents have remained accurate and current - creating an unsustainable bottleneck that caps deployment. Additionally, just over half of respondents (51%) report spending significant staff hours manually reviewing and correcting autonomous AI agent outputs before they go live. This operational friction is felt even more acutely among large enterprises: decision-makers at organizations with $100 million or more in revenue are significantly more likely to report manual review drains (64%) and state that the root cause of project failure almost always traces back to poor data foundations (96%).

In response to these operational hurdles, enterprises are fundamentally restructuring their organizational lines to bring machine intelligence and data governance together. More than half of decision-makers (53%) report that the reporting line for their AI function has moved closer to the primary data organization over the past 12 months, rising to 62% among organizations with $100 million or more in revenue. At the same time, an overwhelming 84% of organizations say they have clearly defined executive accountability when autonomous agents produce flawed or harmful outputs.

As global oversight accelerates, enterprise leaders are also taking proactive measures to prepare for shifting regulatory standards. Nine in ten leaders (90%) report that their organizations are actively preparing for evolving AI regulations across federal, state, and international jurisdictions. To ensure long-term compliance and operational transparency, organizations are primarily focusing on establishing clear internal accountability for AI outputs and decisions (58%) and making targeted investments in data lineage and documentation (51%).

"To move from pilots to enterprise-wide production, businesses cannot simply deploy smarter models; they must build the right context and control directly into their workflows," added Van de Maele. "We have focused our latest innovations on solving this exact bottleneck. By providing organizations with the architecture to govern context and enforce automated guardrails at runtime, leaders can finally eliminate the operational friction holding AI back and scale autonomous workflows with certainty."

For more information and to download the complete The 2026 Hallucination Tax Report, please visit collibra.com.

Research Method
This survey was conducted online within the United States by The Harris Poll on behalf of Collibra from August 5 to 11, 2026, among 306 U.S. adults aged 21+ who are employed full-time as data management, privacy, and/or artificial intelligence (AI) decision-makers (director level or higher) at their current company. Data are weighted where necessary based on company size to bring them in line with their actual proportions in the population. The sampling precision of Harris online polls is measured by using a Bayesian credible interval; for this study, the full sample data is accurate to within +/- 6.4 percentage points using a 95% confidence level.

About Collibra
Collibra is the enterprise AI control plane. Built over eighteen years of ontology engineering, it governs context and control across any data, any model and any agent for AI leaders who need certainty at the speed of AI. A 5x Forrester Wave Leader and 2x Gartner Magic Quadrant Leader, Collibra serves more than 700 customers, including 78 of the Fortune 500 and governs more than 2 billion assets.To learn more, visit collibra.com and follow us on LinkedIn, X, Facebook and Instagram.

SOURCE Collibra