AI trust gap widens as enterprises struggle with data, governance challenges
New report finds most organisations lack real-time data, semantic clarity, and guardrails needed to scale agentic AI
18 April, 2026
TT
16
A growing “trust gap” in enterprise artificial intelligence (AI) is threatening the scalability of next-generation systems, with most organisations struggling to meet the data and governance requirements needed for reliable deployment, according to a new global report.
The AI Trust Gap Report by Denodo, based on a survey of 850 enterprise leaders, finds that while AI is rapidly evolving into “agentic” systems capable of autonomous decision-making, organisations are falling short in three critical areas: access to live data, identifying the right data, and implementing governance guardrails.
The findings come at a time when enterprises are pushing to move AI beyond experimentation into operational workflows, where systems must act in real time and interact directly with business processes.
Real-time data emerges as a critical bottleneck
The report highlights a significant disconnect between enterprise expectations and current capabilities. Around 66 per cent of organisations say AI systems require real-time data to be trustworthy, yet most existing data architectures are not designed to deliver sub-minute latency.
Nearly half of respondents expect data to be available in real time, while only a small minority can rely on historical-only datasets.
This gap is becoming more pronounced as AI systems move into operational roles such as customer service, compliance monitoring, and supply chain optimisation—areas where delayed or outdated data can directly impact outcomes.
Beyond access, identifying and preparing the “right data” continues to be a major challenge. The report finds that 63 per cent of organisations struggle to determine which data is trustworthy or relevant for AI use cases.
Survey insights show that the top data-related challenges include identifying relevant data sources (34 per cent), integrating them (33 per cent), and managing delays caused by data issues (29 per cent).
The problem is compounded by inconsistent business definitions across systems. Variations in how terms such as “customer” or “risk” are defined can lead to conflicting interpretations, undermining AI outputs and decision-making.
Enterprise AI initiatives are also becoming increasingly complex, drawing on an average of more than 400 data sources, with some organisations accessing over 1,000.
This multi-source reality, detailed in the distribution chart on page 12, significantly increases the difficulty of maintaining consistency, governance, and performance across systems—particularly in hybrid and multi-cloud environments.
Governance and security gaps hinder trust
The report identifies governance as another major barrier, with 67 per cent of respondents citing complexity in AI data security and access controls.
Ensuring consistent policy enforcement across distributed systems remains a challenge, particularly as AI agents gain the ability to take autonomous actions. Weak guardrails can expose organisations to risks ranging from compliance breaches to unintended system actions.
The findings align with broader industry concerns, with security frameworks increasingly warning about risks such as “excessive agency” in AI systems.
Even organisations that have invested in modern data platforms are facing operational constraints. Nearly 60 per cent report difficulty optimising performance for AI workloads, highlighting the limits of existing lakehouse and data catalog solutions.
The report notes that agentic AI systems are inherently cost-intensive, often requiring repeated data retrieval, tool execution, and audit logging—raising concerns about long-term economic viability.
The findings reinforce wider industry trends, with the report noting that many organisations are struggling to translate AI adoption into scaled outcomes. Independent estimates suggest that a significant share of agentic AI projects could be abandoned due to rising costs, unclear value, and inadequate risk controls.
To bridge the trust gap, the report calls for a fundamental rethink of enterprise data strategies, emphasising the need for:
- Real-time access to operational data
- Semantic consistency across systems
- Centralised governance and policy enforcement
- Cost and performance optimisation at scale
The study concludes that without these capabilities, enterprises risk stalling AI initiatives at the pilot stage, limiting their ability to deliver measurable business outcomes.



























