The point at which repeated approval requests cause users to stop evaluating each one carefully. In agent governance, this is a control failure mode because the human reviewer becomes desensitised, making the oversight layer ineffective even though the workflow still appears compliant.
What Approval Fatigue Means in Agent Governance
Approval fatigue is a control degradation pattern, not just a user-behaviour issue. It emerges when a reviewer sees enough repeated requests that each new approval starts to feel routine, which weakens judgment and turns a formal gate into a checkbox.
In agent governance, that matters because the human is part of the control itself. If the approver stops scrutinising each request, the workflow may still record a sign-off, but the oversight function no longer meaningfully constrains agent action.
Why Repeated Approvals Break Oversight
The core failure is desensitisation. Repetition lowers attention, especially when requests look similar, arrive at high volume, or are framed as low risk. Over time, reviewers rely on pattern matching and habit rather than case-by-case evaluation, which is exactly what an approval gate is supposed to prevent.
This is why approval design matters as much as approval policy. Bounded approval scopes, clear risk tiers, and separating routine low-impact actions from higher-consequence actions help keep the reviewer cognitively engaged. The problem is not the existence of a human-in-the-loop step, but whether that step still carries real deliberative weight.
How Approval Fatigue Shows Up in Practice
Approval fatigue often appears as faster sign-off times, reduced comment quality, rubber-stamping of recurring requests, or missed anomalies in otherwise familiar-looking actions. It can also show up when teams assume that auditability alone proves effective control, even though the reviewer is no longer applying meaningful scrutiny.
In agentic systems, this is especially dangerous when requests are frequent, partially automated, or presented through polished interfaces that make risky actions seem ordinary. The more an approval flow resembles a normal operational routine, the easier it is for vigilance to drop.
Controls such as Human-in-the-Loop for AI Agents Guide are most effective when the approval is selective, risk-based, and genuinely reviewable rather than universally applied to every action. The same logic is reflected in OWASP Agentic Top 10 ASI09 Guide, which addresses human-agent trust exploitation and the ways overreliance can weaken review quality.
What Good Approval Design Changes
Effective approval design reduces repetition where repetition is harmful and adds friction where the consequence justifies it. That usually means risk-tiered gates, step-up review for unusual or sensitive actions, and approval criteria that force the reviewer to assess substance rather than merely confirm that a request exists.
It also means treating approval quality as a control objective. A control that is technically present but behaviourally ineffective should be monitored like any other degraded safeguard, because apparent compliance can hide real oversight failure.
Risk and Threat Considerations
Approval fatigue creates a material governance and security exposure because it turns a human approval gate into a predictable bypass path. In agent-driven workflows, an attacker or poorly governed automation can exploit that predictability by issuing many benign-looking requests until the reviewer stops applying real scrutiny.
Failure mechanism: Repeated low-friction approvals normalise sign-off behaviour, reduce attention to outliers, and make it more likely that an unsafe agent action is accepted without proper evaluation.
Impact: The organisation can end up with compliant-looking logs but ineffective oversight, allowing unsafe tool use, privilege misuse, or inappropriate actions to pass through the approval layer.
Standards & Framework Alignment
This section maps relevant standards and security frameworks to the operational risks and controls described in this guidance.
OWASP Agentic AI Top 10 addresses the attack surface, NIST SP 800-53 Rev 5 and NIST CSF 2.0 set the technical controls, and EU AI Act defines the regulatory obligations.
| Framework | Control / Reference | Relevance |
|---|---|---|
| OWASP Agentic AI Top 10 | ASI09 — Human-Agent Trust Exploitation | Approval fatigue weakens trust boundaries and human review in agent workflows. |
| Recommendation — Design agent approvals to force meaningful review and prevent routine rubber-stamping. | ||
| NIST SP 800-53 Rev 5 | AC-6 — Least Privilege | Approval fatigue matters where repeated sign-off can let excess access or action persist. |
| AU-6 — Audit Record Review, Analysis, and Reporting | Approval fatigue is detectable when approval evidence shows shallow or perfunctory review patterns. | |
| Recommendation — Limit agent permissions so a weakened approval step cannot authorize broad misuse. Review approval records for repetitive sign-off patterns and missed anomalies. | ||
| NIST CSF 2.0 | PR.AA-05 — Identity Management, Authentication, and Access Control | Approval fatigue affects the effectiveness of access decisions and human review in control paths. |
| Recommendation — Apply risk-based access decisions so approval controls remain substantive. | ||
| EU AI Act | Human oversight for high-risk AI systems | The term directly concerns human oversight quality in AI-controlled workflows. |
| Recommendation — Ensure human oversight remains meaningful rather than nominal in high-risk AI operations. | ||
Practitioner Guidance
What to watch for: Look for approval flows where volume is high, request content is repetitive, and reviewers rarely challenge or reject requests. Those are the conditions where desensitisation is most likely to erode control value.
Governance implication: Approval gates should be owned as active controls, not ceremonial checkpoints. If a gate cannot sustain meaningful human review under realistic workload, it needs redesign rather than additional documentation.
Related resources from NHI Mgmt Group
- How can teams tell whether MFA is causing approval fatigue?
- How should security teams implement push authentication in remote work environments without creating approval fatigue?
- How should security teams automate SaaS access reviews to reduce approval drift and reviewer fatigue?
- What are the signs that MFA is being undermined by prompt fatigue and repeated approval requests?
Deepen Your Knowledge
Free weekly newsletter
Subscribe to the NHI & AI Identity Journal
The latest on NHI and Agentic AI security – articles, research, breaches, news and events every week.
Bonus 33% off our NHI Course when you subscribe.
Reviewed and updated by the NHIMG editorial team on October 30, 2026.
NHI Mgmt Group — the #1 independent authority on Non-Human Identity, IAM, and Agentic AI security. nhimg.org