Login Get a Demo
01 Glossary · AI security

What Is Excessive Agency?

OWASP’s term for an AI system that can do more than its task needs: the three root causes, the fixes, and how it differs from prompt injection.

The short answer

What is excessive agency?

Last updated
Read time
5 min read
From
Cranium AI
Also called
LLM03:2026 Excessive Agency
Share on LinkedIn

Excessive agency is the vulnerability that lets an AI system take damaging actions in response to unexpected, ambiguous or manipulated model output, because it was given more functionality, permission or autonomy than its task needs.

OWASP ranks it third in the Top 10 for LLM Applications 2026, as LLM03:2026, up from sixth in the 2025 edition.

Key takeaways
  • OWASP identifies three root causes: excessive functionality, excessive permissions and excessive autonomy.
  • It is LLM03:2026 in the current OWASP list. In the 2025 edition it was LLM06.
  • Prompt injection is the input-side compromise. Excessive agency is what gives it consequences.
  • OWASP says authorization must be enforced in code, not left to the model’s judgment.
02 Root causes

What causes excessive agency?

OWASP’s LLM03:2026 entry names three, and most over-provisioned agents have all of them.

01

Excessive functionality

Tools include capabilities the task does not need. OWASP’s example is a tool meant only to read documents that can also modify or delete them.

OWASP LLM03:2026
02

Excessive permissions

Tools hold rights on downstream systems beyond what is needed, or act with a broader identity than the user’s, such as a generic high-privilege account that can reach every user’s data.

03

Excessive autonomy

High-impact actions proceed without independent verification or approval, for example deleting records with no confirmation step.

03 Fixes

How do you reduce excessive agency?

The steps follow OWASP’s mitigation list for LLM03:2026.

  1. Minimize tools and functions

    Offer the agent only the tools the task requires, implement only the functions each tool needs, and prefer granular tools with strict input schemas over open-ended ones.

  2. Minimize permissions

    Grant each tool the minimum access to downstream systems, enforced by those systems’ own permissions.

  3. Act as the user

    Execute tools in the requesting user’s context with least privilege, and carry that context through chained and multi-agent calls.

  4. Require approval for high impact

    Put a human in the loop before irreversible actions. Low-consequence, reversible actions can be approved automatically.

  5. Mediate every request

    Validate each downstream request against policy in code: at the tool, at an independent pre-execution policy point or in the downstream system.

  6. Monitor and rate‑limit

    Log tool use and set invocation thresholds. OWASP notes that these limit damage but do not prevent it.

04 Compare

Excessive agency vs. prompt injection

Excessive agencyPrompt injection
What it isToo much capability granted to the systemInput that alters the model’s behavior
OWASP 2026 IDLLM03:2026LLM01:2026
Where the fix livesTool design, permissions and approvalsInput handling, content separation and runtime inspection
Without the otherDormant until the model errs or is manipulatedMostly confined to the conversation
05 Keep going

Related reading.

06 Questions

Frequently asked questions.

Straight answers, kept current. Last updated .

Ask us anything →
What is an example of excessive agency?

OWASP’s first scenario is a hijacked email assistant. A personal assistant is given mailbox access through a tool that can also send messages. A crafted incoming email makes the agent search the inbox for sensitive information and forward it to the attacker’s address. The fixes OWASP lists are a read-only scope, a tool that only reads mail, manual review before sending, or rate limiting on the send interface.

How does excessive agency relate to the OWASP Top 10 for Agentic Applications?

The agentic list breaks the same problem into finer parts. OWASP’s LLM03:2026 entry says that in agentic systems excessive agency can manifest as ASI02 Tool Misuse & Exploitation, ASI03 Identity & Privilege Abuse and ASI08 Cascading Failures. Use LLM03 to reason about one application’s tools and permissions, and the agentic entries when agents plan, delegate and call each other.

Why did excessive agency move up in the 2026 list?

It was sixth in 2025. OWASP’s 2026 document calls its climb to third “the most consequential move on the list” and gives the reason: “the vote and the record agree that agentic deployments are where the damage is landing.” As applications gain tools, a manipulated or mistaken model output stops being a wrong answer and becomes an action.

Can runtime inspection fix excessive agency?

Not by itself. OWASP states that sanitizing model inputs and outputs is not a root control for excessive agency. The root controls are tool design, permissions and approval. Cranium AI is explicit about its own boundary here: AgentSensor inventories agent frameworks, MCP servers and tools from source code, and the Tool Inspector shows the tool calls an agent made. It does not authorize or block tool calls at runtime.

07 See it for yourself

From the definition
to the evidence.

See how Cranium AI discovers, observes, governs, secures and proves every AI model, agent and vendor in your enterprise.