Veeam has presented a new global study at VeeamON in London. Veeam’s new Data and AI Trust Gap report reveals a significant and growing gap at the core of enterprise AI. 88% of organizations already use AI agents, but only 7% meet the requirements to be truly AI ready.
Additionally, 95% say data-related challenges have already slowed their progress in AI. As agentic AI moves from pilots to production environments, organizations face an urgent challenge: ensuring that the data powering those systems is visible, governed, secure, and resilient. In this context, the expansion of agentic AI is redefining data control and monitoring requirements.
The study, based on a global survey of 600 senior managers across the financial, healthcare, manufacturing, retail and technology sectors, reveals that AI adoption is growing at a much faster pace than the governance structures designed to manage it, especially with the rise of agentic AI. Despite heavy investment and executive intent, the ability to control, monitor and recover from AI failures is underdeveloped.
Key findings show that AI is scaling faster than control, a trend even more pronounced in agentic AI initiatives:
- Only 7% of organizations are truly ready for AI.
- 88% already use or are testing AI agents.
- Only 28% are confident they can detect AI systems that operate outside approved parameters, which is critical in agentic AI environments.
- 95% say data challenges have already slowed their AI advancements.
The numbers reveal a clear trust gap between AI adoption and the governance, visibility and control needed to support it, especially in agentic AI deployments.
“Most organizations don’t have a problem with AI adoption, but with trust in it,” said Anand Eswaran, CEO of Veeam. “The first phase of AI was characterized by investment in infrastructure, experimentation and acceleration. The next phase will be defined by trust. With the widespread adoption of autonomous AI agents that operate at machine speed – a hallmark of agentic AI – the question is no longer whether you can use AI, but whether you can ensure that all data is secure, governed, compliant and resilient. And if something goes wrong, can it be accurately recovered? That is the key to accelerating secure AI at scale without simultaneously increasing operational and reputational risks.”
Managers’ confidence hides an operational gap
The study reveals a significant perception gap between senior management and operational teams responsible for AI implementation. Progress often stalls between intent and execution: governance is inconsistent, data is managed reactively, and responsibility is assigned but fragmented, making it difficult to scale agentic AI projects.
- 65% of CEOs believe they have a complete AI inventory, compared to only 48% of technical leaders.
- 52% of CEOs believe they actively lead data management, but only 41% of CISOs and 38% of CIOs agree.
- 48% of CEOs believe that having reliable, secure and compliant data could generate more than 25% revenue growth.
- 83% of CEOs feel pressure to accelerate their AI and data management capabilities.
This combination of rapid AI adoption, coupled with incomplete visibility and unclear responsibilities, creates the perfect conditions for failures that are difficult to detect, explain, and contain.
When AI fails it does not manifest as a service interruption
As AI systems become more autonomous, the nature of failures is changing, especially in agentic AI scenarios. Risk is shifting from traditional system outages to data-level failures that are harder to detect, explain, and contain. The study warns that errors produced at machine speed can exceed detection capacity, forcing resilience to evolve from general to precise recovery: restoring only what was affected, instead of reverting entire environments.
Among organizations currently using AI, only a minority were able to identify within minutes:
- What systems were accessed (29%).
- What actions did you take (25%).
- What decisions did it influence (24%).
- What data the system used (22%).
Additionally, only 40% of leaders are fully confident that they can accurately isolate and reverse a failure caused by an AI agent.
From the inside out and from the outside in: governance converges on data
The governance challenge converges on data from two perspectives: internal demand and external regulatory pressure, both intensified by the growing use of agentic AI.
Within organizations, unauthorized use of AI is now commonplace:
- 95% report unauthorized use of AI within their organization and 93% consider it a significant risk.
- However, only 25% offer approved alternatives, meaning most are trying to suppress demand rather than manage it effectively.
- 44% point to the increase in cyber risk as the main threat associated with “Shadow AI”.
At the same time, regulatory pressure external to the organization is intensifying. 61% of organizations say the EU AI Law has already influenced AI investment strategies in the last 12 months, while 47% indicate that maintaining audit trails for AI decisions is their biggest compliance concern.
Trust requires shared responsibility, not ambiguity
The new study shows that the main barriers to progress are fragmentation of responsibilities and lack of operational alignment, where data, AI and governance responsibilities are often dispersed across teams, diluting accountability and slowing execution. When “everyone is accountable,” no one can decisively set policies, enforce controls, or demonstrate results.
When responsibility is clearly defined, results improve significantly:
- Organizations where CISOs are responsible for risk associated with AI agents are 24% more likely to detect irregular behavior.
- Organizations that rely on shared responsibility models are 47% less likely to detect these behaviors.
- Data doesn’t need another champion, it needs responsible leadership strong enough to align governance, security, privacy, compliance and resilience.
Trust is becoming the operating foundation of enterprise AI
There is a clear difference between organizations that can operate with confidence and those that cannot. Organizations that successfully align ambition, visibility and governance significantly outperform others.
Among organizations classified as fully AI ready, 97% report having achieved measurable business benefits from their investments in data and AI, compared to 48% overall, demonstrating the value of turning trust into an enterprise-scale operational capability.
Veeam: creating the trust layer for data and AI
Veeam addresses this challenge by combining resiliency, security, and data governance to help organizations visualize what data AI uses, control how people and agents access it, and accurately recover clean, trusted data when incidents occur—a key aspect to the success of agentic AI.
The study shows that the main barriers to progress are the fragmentation of responsibilities and the lack of operational alignment
“The results are unequivocal. When 95% of executives say that data-related challenges are already slowing their progress in AI, the bottleneck is not the model, but trusted, governed and recoverable data,” Eswaran added. “Veeam is building the trust layer for data and AI to give enterprises the visibility, control and precise recovery needed to safely scale AI and drive real business value.”
