TL;DR: AI use is widespread while only 13% of organisations report strong visibility into how it touches data, according to Cyera's 2025 State of AI Data Security Report. That leaves prompt-layer risk and over-permissioned access hard to govern before they affect sensitive data.
Editorial analysis by NHI Mgmt Group, based on content published by Cyera: “83% Use AI; Only 13% Have Visibility - Cyera’s 2025 State of AI Data Security Report”.
By the numbers:
- 83% of enterprises already use AI in daily work, according to Cyera research.
- Only 13% report strong visibility into how AI touches enterprise data, according to Cyera research.
- Only 11% can automatically block risky AI activity, according to Cyera research.
Key questions
Q: Why do AI agents create a bigger governance problem than ordinary endpoint tools?
A: Because an AI agent can execute many file reads, API calls, and transfers in one session without a human approving each step.
Q: How can organisations tell whether their AI security model is actually working?
A: They should test whether the control stack can explain who acted, what data was touched, and what purpose the action served.
Q: What breaks when AI agents are treated like standard human users?
A: You lose visibility into effective permissions, expected behaviour, and real blast radius.
Practitioner guidance
- Instrument AI activity at the prompt layer Capture prompts, outputs, access decisions, and anomaly signals early enough to support containment rather than only post-incident review.
- Assign AI a first-class identity model Treat AI systems as governed identities with named owners, scoped permissions, and explicit lifecycle control rather than as informal extensions of user access.
- Constrain autonomous agent scopes Limit the data and tool reach of autonomous agents to the smallest task-specific boundary, and require approval gates for higher-risk actions.
Bottom line: AI use is already widespread, but governance still trails the way AI touches data in practice.
Glossary terms: Interaction-Layer Visibility, Autonomous AI Agent, First-Class Identity Object, Over-Permissioning
Explore further
View Full Forum → | NHI Foundation Course → | Our Services → | Read the full analysis →
AI governance is becoming an identity problem before it becomes a data problem. Cyera’s survey shows that enterprises are already using AI at scale while only a small minority can see how it touches data in real time. That means the first failure is not exfiltration, but the absence of a governable identity boundary around AI activity. Practitioners should treat AI access as a first-class entitlement model, not a side effect of application rollout.
A few things that frame the scale:
- 53% of security leaders expect AI to run major portions of their infrastructure autonomously within the next three years, according to the 2026 Infrastructure Identity Survey.
- 19% of organisations give AI systems dramatically more access than human employees, nearly one in five granting unrestricted privilege, according to the 2026 Infrastructure Identity Survey.
A question worth separating out:
Q: Should organisations use the same access model for humans and AI agents?
A: No. Human access models are built around stable roles and review cycles, while AI agents often need contextual, task-specific permissions that change quickly. Treating them the same usually leads to over-permissioning or constant exceptions. Organisations should separate identity proof from authorization design and apply resource-level controls for agents.
👉 Read our full editorial: AI security needs agent-specific controls, not human-style governance