TL;DR: Unauthorized AI use is widespread in the workplace, with WatchGuard reporting that 64% of employees admit using unapproved AI tools and fewer than 30% believe their organisation has an accurate software inventory, which points to a broader SaaS visibility gap. Shadow AI is exposing a continuous discovery problem, not just an AI policy problem.
At a glance
What this is: NHI Mgmt Group's analysis, drawing on WatchGuard's report: Shadow AI is revealing a wider SaaS visibility gap, with 64% of employees admitting to unauthorized AI use and fewer than 30% believing their organisation has an accurate software inventory.
Why it matters: IAM and security teams need continuous visibility into SaaS applications, integrations and permissions because unmanaged AI use often appears first as an access and inventory problem, not a standalone AI policy breach.
By the numbers:
- 64% of employees admit to using unauthorized AI tools for work.
- Fewer than 30% of respondents believe their organisation maintains an accurate software inventory.
- 39% say their organisation does not maintain an accurate software inventory.
Context
Shadow AI is best understood as a visibility and governance problem, not only a policy problem. Employees can introduce AI services into workflows without approval, and those services may connect to corporate data through SaaS permissions, OAuth grants and embedded integrations that security teams never explicitly reviewed.
The control gap is broader than unknown applications. Even approved SaaS can become risky when permissions expand, integrations are added or identity activity changes faster than point-in-time reviews can capture. That makes continuous discovery the relevant baseline for governance, not periodic inventory checks.
Key questions
Q: Why do shadow AI tools create more risk than sanctioned SaaS apps?
A: Shadow AI bypasses procurement, security review, and entitlement design, so it often enters with broad access and no clear accountability. Even when the tool is well intended, the absence of an owner and review cadence means the organisation cannot reliably enforce data handling, access control, or revocation.
Q: Why do SaaS inventories miss shadow AI usage?
A: SaaS inventories miss shadow AI usage because users sign up for external AI tools outside approved procurement, often with personal email accounts or browser sessions that never touch corporate CASB, SSO, or SaaS discovery controls. The result is unsanctioned data sharing, hidden API tokens, and unmanaged access paths that bypass asset registers and governance reviews.
Q: What is the difference between OAuth and direct login?
A: OAuth delegates authorisation to a trusted application so it can obtain scoped access tokens without learning the user’s password. Direct login authenticates the user directly to the application with their credentials, often via username and password plus MFA. OAuth is for delegated access, while direct login is for primary authentication and account sign-in.
Q: What should organisations do first when shadow AI appears in the environment?
A: Organisations should treat the tool as an unmanaged identity, then decide whether it is allowed to handle business data at all. The first 24 to 72 hours should focus on discovery, containment, and policy enforcement at the access boundary. If the tool cannot be governed inline, it should be denied by default.
Technical breakdown
Why OAuth-connected SaaS creates hidden access paths
OAuth lets one application obtain delegated access to another system using scoped tokens rather than shared passwords. That design is useful, but it also means a legitimate app can reach email, files or calendars without appearing as a traditional login event. The risk grows when grants are broad, long-lived or never revalidated against business need. In practice, the exposure is not only the app itself but the permission relationship created between user, tenant and integration. See the Shadow AI and AI Agent Discovery Guide.
Practical implication: Review delegated grants as privileged access paths and remove SaaS integrations that no longer need the scopes they were granted.
Why point-in-time inventory misses SaaS risk
A software inventory captures what existed at a moment in time, but SaaS environments change continuously. Users add applications, admins approve integrations, vendors modify features and permissions drift as teams adopt new workflows. The result is a moving target that can look controlled during an audit window while still containing unmanaged applications and unexpected access paths. Security leaders should treat SaaS discovery as an ongoing control process rather than a periodic hygiene task.
Practical implication: Shift from periodic inventory reviews to continuous discovery of applications, integrations and identity activity.
How shadow AI becomes an identity governance issue
Shadow AI rarely remains an isolated tool choice. Once employees use it for work, the service may inherit corporate identities, SaaS sessions or OAuth authorisations, which places it inside the same governance chain as other business applications. That makes identity, configuration and usage telemetry part of the same control surface. The key challenge is not merely whether a tool is approved, but whether the organisation can see who connected it, what it can reach and whether that access still makes sense. See the Insider Risk from AI Tools Guide.
Practical implication: Bring AI services into the same governance model used for SaaS identities, permissions and access review.
Threat narrative
Attacker objective: The objective is to reach corporate data through unmanaged SaaS and AI integrations that bypass normal governance and review.
- Entry occurs when employees adopt unauthorized AI services and connect them to work processes without IT review.
- Credential and permission exposure follows when those services obtain OAuth grants, SaaS sessions or other delegated access to corporate resources.
- Impact emerges when unmanaged integrations or excessive permissions expose email, files or other business data outside the organisation's visibility.
Breaches seen in the wild
- Vercel Context.ai OAuth Supply Chain Breach: Shadow AI app Context.ai OAuth integration exposes Vercel customer data via unmanaged third-party token.
Read and download The State of NHI & AI Agent Breach Report 2026, covering 200+ breaches impacting Non-Human Identities including AI Agents.
NHI Mgmt Group analysis
Shadow AI is a symptom of SaaS governance collapse, not the root cause. The real control failure is that organisations cannot reliably see their application estate, its integrations or the identities attached to it. When visibility is incomplete, policy enforcement becomes selective and review becomes retrospective. The practical conclusion is that shadow AI should be treated as an inventory and access governance failure first, and an AI policy problem second.
Continuous visibility matters more than periodic hygiene checks. SaaS changes faster than quarterly reviews, so the control model must follow the environment rather than the calendar. Applications, permissions and integrations drift continuously, which makes stale assessments a poor basis for trust. Organisations that keep relying on snapshot inventories will keep discovering risk after users have already normalised it.
OAuth grants are the new hidden access perimeter. Many teams still think about SaaS risk in terms of approved applications, but delegated access is where the operational exposure lives. An approved service can still reach sensitive resources in ways that never show up as a conventional login or direct admin action. That makes delegated authorisation part of identity governance, not just application configuration.
Visibility debt is the named concept this pattern exposes. When organisations tolerate incomplete software inventory, they accumulate a gap between what is deployed and what is governed. Shadow AI makes that debt easier to see because employees adopt tools quickly, but the underlying issue predates AI and will outlast it. The better control objective is to reduce the gap between real usage and governed usage before it turns into an access incident.
From our research library:
- 8 of the top 10 fastest-growing types of leaked secrets year-over-year are tied directly to AI services, according to the State of Secrets Sprawl 2026.
- Only 5.7% of organisations have full visibility into their service accounts, according to the Ultimate Guide to NHIs.
- Read next: Shadow AI and AI Agent Discovery Guide
What this signals
Shadow AI makes access drift visible: when employees can introduce tools faster than security can classify them, the gap is no longer theoretical. The practical issue is not just discovering an app once; it is maintaining an accurate map of what that app can reach as permissions and integrations change.
Visibility debt is now a governance metric, not a background condition. Organisations that cannot account for applications, identities and delegated access will keep absorbing risk from the next wave of embedded AI, SaaS extensions and unmanaged integrations.
Continuous discovery should become the default control pattern for SaaS estates. Snapshot inventory will still matter for audits, but operational security depends on tracking authorised connections and unauthorised adoption in near real time.
For practitioners
- Implement continuous SaaS discovery Track applications, integrations, permissions and identity activity as a living control, not a quarterly audit output.
- Review delegated OAuth grants Classify SaaS integrations by scope, business owner and last-use date, then remove broad or orphaned authorisations.
- Unify AI and SaaS governance Bring unauthorized AI services into the same approval, inventory and access review process used for other business applications.
- Measure inventory accuracy as a control metric Track the difference between approved software, observed software and connected identities so visibility gaps are visible to operations and leadership.
Key takeaways
- Shadow AI is exposing a wider SaaS governance gap, where inventory accuracy and delegated access matter more than isolated tool approval.
- The source report ties the problem to concrete visibility shortfalls, including widespread unauthorised AI use and limited confidence in software inventories.
- Security teams need continuous discovery, delegated access review and identity-aware SaaS governance to reduce exposure before it becomes an incident.
Related NHI Mgmt Group resources: Shadow AI and AI Agent Discovery Guide · Top 10 Shadow AI Risks · Insider Risk from AI Tools Guide · AI Usage Monitoring Guide
Standards & Framework Alignment
This section maps relevant standards and security frameworks to the operational risks and controls described in this guidance.
OWASP Non-Human Identity Top 10 addresses the attack and risk surface, while NIST CSF 2.0 and CIS Controls v8 set the governance and control requirements practitioners need to meet.
| Framework | Control / Reference | Relevance |
|---|---|---|
| OWASP Non-Human Identity Top 10 | NHI-03 — Vulnerable Third-Party NHI | Shadow AI often arrives through third-party SaaS and integrations that are not governed as trusted identities. |
| NHI-09 — NHI Reuse | The same user or service identity often gets reused across many SaaS and AI connections, hiding risk. | |
| Recommendation — Inventory third-party SaaS identities and remove any integration that lacks an owner or approved business scope. Map reused identities across SaaS connections and separate business-critical access from opportunistic tool usage. | ||
| NIST CSF 2.0 | ID.AM-01 — Physical devices and systems within the organisation are inventoried | A missing or stale software inventory is the core governance gap behind shadow AI exposure. |
| PR.AA-05 — Access Permissions, Entitlements and Authorizations | OAuth grants and SaaS permissions define what shadow AI can reach. | |
| Recommendation — Maintain a current inventory of applications and connected services so unmanaged SaaS can be identified quickly. Review and limit delegated SaaS permissions so access remains aligned to actual business need. | ||
| CIS Controls v8 | CIS-5 — Account Management | Shadow AI becomes risky when accounts and authorisations are created outside managed lifecycle controls. |
| Recommendation — Track SaaS accounts and authorised integrations under a single account-management process. | ||
Key terms
- Shadow AI: AI agents, copilots, or connected tools operating without full visibility or governance from security teams. Shadow AI becomes an identity problem when those systems authenticate with unmanaged tokens, service accounts, or OAuth apps that can reach production resources.
- OAuth Grant: An OAuth grant is the delegated permission an application receives to act on a user's behalf without storing the user's password. In NHI governance, it should be treated as a standing identity relationship with scope, ownership, and revocation requirements, not as a one-time setup detail.
- Delegated Access: Delegated access is permission granted to one identity to act on behalf of another user, service, or system. In NHI environments, this usually appears in OAuth-connected apps and automation tooling. It is powerful, but it must be tightly scoped and reviewed because it can persist long after the original business need ends.
- SaaS Visibility: SaaS visibility is the ability to identify which software services, tenants, and accounts exist in an environment and who controls them. In identity governance, it is the prerequisite for review, offboarding, and cost control because hidden applications cannot be certified or revoked reliably.
Deepen your knowledge
The NHI Foundation Level course, the industry's only accredited NHI security programme, covers NHI governance, machine identity security and secrets management. It helps practitioners connect identity controls to the broader access risks that appear when new tools enter the environment.
Published by the NHIMG editorial team on November 9, 2026.
NHI Mgmt Group, the independent authority on Non-Human Identity, IAM, and Agentic AI security. nhimg.org