Back to Blog
AI Security

OWASP LLM06 Excessive Agency: What It Is and How to Prevent It

The #1 agentic AI risk — and the three controls that prevent LLM06 violations without eliminating agent autonomy

OWASP LLM06 is the vulnerability that amplifies every other LLM risk. System prompt instructions do not prevent it — capability-level controls do. Here are the three technical controls: minimum-capability tool scoping, confidence-based escalation thresholds, and human-in-the-loop enforcement for irreversible actions.

July 30, 202611 min read
AI SecurityOWASP LLM06Excessive AgencyPrompt InjectionTool ScopingHITL
AgentTrustOSAGENTIC AI GOVERNANCESECURITY · OWASP LLM TOP 10OWASP LLM06:Excessive Agency — What It Is,Why It Spreads, How to Stop ItDENIED SCOPEREQUESTEDPERMITTEDSCOPE!OWASP LLM06 ViolationSources: OWASP LLM Top 10 2025 · NIST AI RMF 2.0 · Gartner Agentic AI Risk Report 2025agent-trust.tech
Permission scope violation — the gap between permitted scope and what an agent requests or takes is the Excessive Agency attack surface
Key Facts
  • According to OWASP's Top 10 for Large Language Models (2025 edition), LLM06 Excessive Agency ranks as the primary risk when agents take consequential actions with insufficient oversight or scope constraints.
  • The OWASP LLM06 taxonomy defines Excessive Agency across three dimensions: excessive functionality (too many tools), excessive permissions (over-broad scopes), and excessive autonomy (insufficient human oversight).
  • Per the OWASP Top 10 for LLM Applications, the recommended mitigations include least-privilege tool allowlisting, time-bounded credentials, and mandatory human-in-the-loop checkpoints for high-impact operations.
  • Microsoft Entra Agent ID and AWS AgentCore Identity both provide per-agent OAuth 2.1 credential issuance as production-ready infrastructure, available today without custom implementation.
  • Escalation thresholds — for example, $50 auto-execute vs. $5,000 require human approval — are the operational translation of OWASP LLM06's "human-in-the-loop" requirement for financial operations.
  • Immutable audit trails, required by frameworks including SOC 2 Type II and ISO 27001, provide the forensic reconstruction capability needed when an Excessive Agency incident occurs.
TL;DR
  • OWASP LLM06 Excessive Agency means agents have more tools, permissions, and autonomy than any task requires — creating unbounded blast radius.
  • Most enterprise agents today run on shared API keys with 40+ tools enabled, no credential rotation, and no human escalation path.
  • The fix is operational, not architectural: per-agent OAuth 2.1 identity, least-privilege tool allowlists, and two-tier escalation thresholds.
  • Microsoft Entra Agent ID and AWS AgentCore Identity make per-agent identity a production reality without custom infrastructure.
  • Immutable audit logging is not compliance theater — it's the forensic record you need to reconstruct what an agent did when an incident occurs.
Keep reading → Full implementation guidance below.

Consider what you know about your most trusted employee. They have access to the systems their role requires. Their badge expires when they leave. Someone approved their access, and someone else can revoke it. If they make a mistake, there's a record. Now consider your average enterprise AI agent. It was provisioned with a shared API key that never rotates. It has access to 40 or more tools — many of which were added during development and never removed. It runs continuously, 24 hours a day, with no badge-out mechanism. Nobody approved its current scope formally, and nobody knows exactly what it can do. This is OWASP LLM06 Excessive Agency.

The cost of inaction is not theoretical. When an agent operating with Excessive Agency is manipulated via prompt injection — a technique where malicious content in the environment overrides the agent's instructions — the result is a fully credentialed actor executing arbitrary actions at scale. The blast radius is determined not by the attacker's capability, but by the agent's provisioned access.

This post covers what OWASP LLM06 Excessive Agency is, why enterprises systematically over-provision agents, how to implement per-agent identity and least-privilege controls in production, and what escalation thresholds actually look like in regulated industry deployments.

THE VULNERABILITY

What exactly does OWASP LLM06 Excessive Agency cover?

OWASP LLM06 Excessive Agency is the sixth entry in the OWASP Top 10 for Large Language Model Applications and specifically addresses the risk that arises when an LLM-based agent is granted more capability than it needs to perform its intended function. Unlike many security vulnerabilities that require an active exploit, Excessive Agency is a design-time failure — the over-provisioning happens during development and persists silently until it matters.

Definition — OWASP LLM06 Excessive Agency

OWASP LLM06 Excessive Agency describes a vulnerability class in which an AI agent is granted excessive functionality (access to tools it doesn't need), excessive permissions (overly broad scopes within those tools), or excessive autonomy (the ability to take high-impact actions without human approval). Any of these dimensions can independently create an unacceptable blast radius if the agent is compromised, manipulated, or simply makes an error.

The three sub-dimensions of LLM06 are worth understanding separately because they have different mitigations. Excessive functionality is addressed by tool allowlisting — defining the minimal set of tools an agent needs for its specific task and stripping all others. Excessive permissions is addressed by scope restriction within each tool. Excessive autonomy is addressed by human-in-the-loop escalation thresholds.

Why do enterprise agents end up over-provisioned by default?

There are four structural reasons that enterprise agents accumulate excessive permissions, and they all trace to the development lifecycle, not to malicious intent.

First, provisioning is easier than scoping. When a developer creates an agent prototype, they provision a service account or API key with broad access to unblock development velocity. Production deployment inherits the prototype's permissions because nobody explicitly reverts them. Second, tool sets are additive. Developers add tools as they identify capabilities the agent needs, but rarely remove tools when requirements change. Third, shared credentials are operationally convenient. A shared API key used across multiple agent instances is simpler to manage than per-agent OAuth tokens — until an incident occurs. Fourth, escalation paths are awkward to implement without a framework.

Definition — Blast Radius (Agentic AI)

In agentic AI security, blast radius refers to the scope of systems, data, and operations that could be affected if an agent is compromised, manipulated, or makes an error. An agent's blast radius is determined by the union of all tools it can access, the permissions within each tool, and the volume of actions it can take autonomously without human oversight. Reducing blast radius is the primary operational goal of OWASP LLM06 mitigations.

PER-AGENT IDENTITY

How does per-agent identity reduce OWASP LLM06 risk?

Per-agent identity is the foundational control for OWASP LLM06. Instead of all agents sharing a single service account or API key, each agent instance receives its own credential that is scoped to the specific task it is performing and expires when that task is complete. This has two security effects: it eliminates credential reuse across agents, and it makes audit trails interpretable — every action in the log is attributed to a specific agent identity, not to an anonymous shared key.

Microsoft Entra Agent ID, released in 2025, provides exactly this pattern for Microsoft ecosystem deployments. Each agent is registered as an identity in Entra ID and receives OAuth 2.1 tokens scoped to specific API audiences and time-bounded to the session. AWS AgentCore Identity provides the equivalent on the AWS side — each AgentCore agent receives an IAM credential scoped to its Cedar Policy-defined permissions. Both implementations support short-lived tokens with automatic rotation.

What escalation thresholds should enterprises configure?

Escalation thresholds are the operational mechanism for implementing OWASP LLM06's human-in-the-loop requirement. They define a boundary between actions the agent may take autonomously and actions that require a human to explicitly approve. Thresholds can be based on financial value, data sensitivity, operational reversibility, or regulatory classification.

In financial services, a common starting pattern uses two tiers: transactions below $50 auto-execute, transactions between $50 and $5,000 require human approval within a defined window, and transactions above $5,000 require explicit sign-off before the agent can proceed. The same logic applies to non-financial operations: deleting fewer than 100 records from a non-regulated table may be auto-approved, but deleting records from a PII-containing table always requires human confirmation.

Definition — Human-in-the-Loop Escalation

Human-in-the-loop (HITL) escalation in agentic AI refers to a configured mechanism that pauses an agent's autonomous execution and routes a pending action to a human approver before proceeding. OWASP LLM06 requires HITL for high-impact operations. Effective HITL implementations define explicit thresholds (financial, data sensitivity, regulatory scope), a designated approver with SLA expectations, a timeout behavior (reject vs. pause), and an audit record of the approval decision and the approver's identity.

The approver role matters. HITL is weakest when the approval goes to the same team that deployed the agent — they are likely to approve without sufficient scrutiny. Effective HITL escalation routes to a named individual in a different chain of command: a compliance officer, a senior engineer who did not develop the agent, or a risk manager.

BY THE NUMBERS

The Excessive Agency gap in enterprise deployments

#6
OWASP LLM06 rank in the OWASP Top 10 for LLM Applications
OWASP, 2025 edition
40+
Average tools enabled per enterprise pilot agent, per security audits
AgentTrust OS customer analysis, 2026
0
Average credential rotation cycles for shared API keys in agent deployments
AgentTrust OS customer analysis, 2026
LEAST-PRIVILEGE ENFORCEMENT

How does least-privilege tool allowlisting work in practice?

Least-privilege tool allowlisting means defining the precise set of tools an agent is permitted to invoke for a given task, and configuring the agent runtime to block any tool call outside that set. This is a departure from the common pattern of giving agents access to all available tools and trusting the model to select appropriate ones.

Implementation requires a tool inventory for each agent — a documented list of every tool the agent can invoke, the permission scope within each tool, and the business justification for that access. For an agent that processes customer refunds, the allowlist might include: read access to the order management system, write access to the refund processing API (restricted to amounts below $500), and read access to the customer record system (name and email only, not full profile).

The allowlist must be enforced at the runtime level, not just at the model level. Models can be instructed not to use certain tools, but a prompt injection attack may override that instruction. Runtime enforcement means the agent execution environment intercepts tool calls and rejects any call to a tool not on the allowlist, regardless of what the model requested.

Agent Action Request — OWASP LLM06 Decision FlowAgent Tool Call Requeste.g. processRefund($2,400)① Identity CheckDoes this agent have a valid, non-expired token?FAILBLOCKEDLog + Alert② Scope CheckIs this tool on the agent's allowlist + scope?FAILBLOCKEDLLM06 Violation③ Threshold CheckDoes action exceed escalation threshold?≤ $50AUTO-EXECUTELog action + result> $50ESCALATEHuman approvalImmutable Audit LogWho · What · When · Why

Figure 1: OWASP LLM06 Excessive Agency — Agent Action Request Decision Flow. Every tool call passes through identity, scope, and threshold checks before execution. Failures are blocked and logged. Escalations route to human approval.

AUDIT LOGGING

What role does immutable audit logging play in OWASP LLM06 defense?

Immutable audit logging serves two functions in an OWASP LLM06 defense posture: it provides the forensic record needed for incident reconstruction, and it creates the accountability mechanism that makes human-in-the-loop escalation meaningful. An approval without a record of who approved it and when is just a delay mechanism, not a governance control.

An effective audit log for agentic AI captures the agent identity (not just the service account, but the specific agent instance and session), the tool called, the parameters passed to that tool, the result returned, the decision path taken (auto-execute, human escalation, or blocked), and — for escalated actions — the approver identity, approval timestamp, and any notes. This record must be append-only and stored in a system the agent itself cannot modify.

In regulated industries, audit logs for agentic AI actions are increasingly expected to meet the same standards as human transaction records. SOC 2 Type II requires evidence that access control policies are operational; an immutable log of agent decisions is the most direct evidence available. ISO 27001 Annex A.8.15 requires logging and monitoring of user activities, and AI agents are being treated as a category of "user" by many compliance frameworks.

HOW THIS GETS SOLVED

AgentTrust OS: implementing OWASP LLM06 controls in production

AgentTrust OS provides the three-layer control stack that directly addresses each dimension of OWASP LLM06 Excessive Agency — pre-production privilege review, real-time enforcement, and immutable audit.

Trust Runtime

Real-time identity and scope enforcement

Validates agent token, checks tool allowlist, applies escalation thresholds, and blocks out-of-scope calls before execution on every tool call.

📋
Trust Audit

Append-only, tamper-evident logging

Immutable log of every agent action, decision, and escalation event. Provides the forensic reconstruction capability required by SOC 2, ISO 27001, and emerging AI audit standards.

🏷️
Trust Certify

Pre-production privilege review

Validates that agent tool allowlists conform to least-privilege policy, flags excessive permissions before agents reach production, and certifies the agent's privilege posture.

Trust CertifyPre-production privilegereview + certificationTrust RuntimeReal-time enforcementIdentity · Scope · ThresholdTrust AuditImmutable log · Forensicreconstruction · Compliance

Figure 2: AgentTrust OS governance pipeline for OWASP LLM06 Excessive Agency — pre-production certification, real-time runtime enforcement, and immutable audit logging.

FREQUENTLY ASKED QUESTIONS

Your questions, answered directly

Prompt injection (OWASP LLM01) is the attack vector — malicious content in the environment that overrides the agent's instructions. Excessive Agency (OWASP LLM06) is the vulnerability that determines how bad the outcome is when prompt injection succeeds. An agent with minimal tool access and strict escalation thresholds can be successfully prompt-injected yet cause minimal harm; an over-privileged agent with the same injection can execute enterprise-wide destructive actions. Both risks must be addressed, but LLM06 determines blast radius.
No. Microsoft Entra Agent ID integrates with existing Entra ID tenants and issues per-agent OAuth 2.1 tokens without additional infrastructure. AWS AgentCore Identity integrates with existing IAM deployments. Both are available as production services today. Custom identity provider development is unnecessary for organizations already using Microsoft 365 or AWS.
Non-financial thresholds should be based on data sensitivity, operational reversibility, and regulatory scope. A useful starting framework: if the action is irreversible (file deletion, record purge, contract execution), always escalate regardless of volume. If the action involves personal data of more than 100 individuals, escalate. If the action would trigger a regulatory notification obligation if it went wrong, escalate. Define these as explicit policy statements in the agent's governance configuration, not as informal guidelines.
Retention requirements depend on the regulatory context. SOC 2 generally requires a minimum of one year, with many organizations retaining two to three years to support incident investigations. GDPR's data minimization principle creates a tension with long retention periods for logs containing personal data — the recommended approach is to log action metadata (agent ID, tool, timestamp, decision) and separately log data content with appropriate access controls and shorter retention periods.
OWASP LLM06 is a specific risk taxonomy; the NIST AI RMF is a governance process. The NIST AI RMF's Govern function requires organizational accountability for AI risk; Map requires identifying specific risks like LLM06; Measure requires assessing the actual exposure; Manage requires implementing controls. OWASP LLM06 mitigations (per-agent identity, least privilege, escalation thresholds, audit logging) are the 'Manage' activities for the 'Map'-identified risk of Excessive Agency.
CONTAIN YOUR BLAST RADIUS TODAY

Implement OWASP LLM06 controls without replacing your stack

AgentTrust OS implements OWASP LLM06 controls across pre-production, runtime, and audit — identity enforcement, least-privilege allowlisting, escalation thresholds, and immutable logging.

Start Free →

More from the blog

AI ComplianceJuly 22, 2026AI ComplianceJuly 22, 2026AI GovernanceJuly 22, 2026AI GovernanceJuly 22, 2026AI ArchitectureJuly 22, 2026AI SecurityJuly 16, 2026MLOpsJuly 10, 2026AI ImplementationJuly 8, 2026AI Agent ArchitectureJuly 5, 2026AI Agent ArchitectureJune 30, 2026EngineeringJune 23, 2026AI Agent ArchitectureJuly 2, 2026AI Agent ArchitectureJuly 1, 2026IntegrationsJuly 1, 2026IntegrationsJuly 1, 2026IntegrationsJuly 2, 2026AI SecurityJuly 3, 2026AI SecurityJuly 1, 2026AI ComplianceJuly 2, 2026AI ComplianceJuly 3, 2026AI StrategyJuly 2, 2026AI StrategyJuly 3, 2026AI StrategyJuly 3, 2026AI StrategyJuly 3, 2026AI GovernanceJuly 28, 2026Healthcare AIJuly 28, 2026ArchitectureJuly 29, 2026ArchitectureJuly 29, 2026AI StrategyJuly 29, 2026AI StrategyJuly 30, 2026AI ComplianceJuly 30, 2026EngineeringJuly 30, 2026AI GovernanceAugust 4, 2026EngineeringAugust 4, 2026EngineeringAugust 4, 2026