Skip to content
IAM Insights

Explainers · Glossary

Confused deputy

confused deputy attack · confused deputy attacks · confused deputy problem

A classic security flaw where a system holding legitimate elevated privileges is tricked by a less-privileged or malicious party into misusing those privileges on that party's behalf; applied to AI agents that hold real permissions but can be manipulated through their prompts or inputs.

In our briefings

The claim each of the newest briefings on Confused deputy would have you leave with.

  1. OWASP's AIVSS v0.8 starts from a CVSS v4.0 base score and adds an uplift worth part of the gap between that base and 10. Ten agentic factors, each scored 0, 0.5 or 1, set the share, and an exploit-maturity multiplier scales it, by default 0.97. With the default mitigation factor of 1.0 (no or weak mitigation), all ten worked examples in the document land between 7.1 and 10.0, seven Critical and three High, and a goal-manipulation flaw with a CVSS base of 2.1 comes out at 7.1. The authors read a jump like that as a sign that the agentic deployment context, not the technical flaw, is the main risk driver. On our reading of the formula, a score can only fall below its CVSS base through the mitigation factor, and the document warns that AIVSS scores are not directly comparable to CVSS base scores.

    OWASP's AIVSS lifts a CVSS 2.1 agent flaw to 7.1, and all ten worked examples land High or CriticalBI-FIND-2026-2299

  2. The draft moves the description of what is being approved from the agent to the API. When a call needs approval, the protected resource normally answers 401 with a challenge it signs itself: a JWT carrying a transaction id and the operation as Rich Authorization Request details. The agent must relay that challenge unmodified and "MUST NOT replace it with an agent-generated description". The authorization server gets a decision from an approving party, who need not be the agent's user, and issues a token bound to that one transaction, which may be narrowed but never broadened and must have a limited lifetime. Since revision -01 the token is tied to a key when the resource names one in the challenge; otherwise the server may tie it to a key the client proves.

    An OAuth draft lets the API, not the agent, describe the approval, and now name the key the token is bound toBI-FIND-2026-0626

  3. WorkOS's checklist limits a delegated agent to the intersection of what the agent is allowed to do and what the user it acts for is "currently allowed to do," and says to check both on every tool call, not once per session. Its example code reads the agent's ceiling from the token and makes a live check for the user, and its test is that revoking a user's access to a project mid-task makes the agent's next call on that project fail. In the post's own example, an agent whose ceiling includes projects:delete still cannot delete a project the user cannot.

    WorkOS's agent checklist limits a delegated agent to what it and its user may both do, checked on every callBI-FIND-2026-1997

›Cited in 17 briefings

Every briefing that discusses this term, newest first.

  1. OWASP's AIVSS lifts a CVSS 2.1 agent flaw to 7.1, and all ten worked examples land High or Critical

    Article · · OWASP AIVSS project · Ken Huang et al. · Oct 22, 2026

  2. An OAuth draft lets the API, not the agent, describe the approval, and now name the key the token is bound to

    Draft · · IETF Datatracker · Yaroslav Rosomakho et al. · Jun 25, 2026

  3. WorkOS's agent checklist limits a delegated agent to what it and its user may both do, checked on every call

    Article · · WorkOS Blog · Maria Paktiti · Sep 30, 2026

  4. Sweeney's delegation draft concedes its server is trusted with users' credentials, and drops an interop claim

    Draft · · IETF Datatracker · Kieran Sweeney · Oct 13, 2026

  5. Niyikiza's agent tokens narrow to a tool's arguments offline; revision -02 breaks -01 implementations

    Draft · · IETF Datatracker · Niki Aimable Niyikiza · Jun 15, 2026

  6. Anthropic asks for per-agent identity; its guide leaves Claude Code sub-agent credentials unspecified

    Article · · Anthropic (Claude) · Anthropic · May 27, 2026

  7. GitHub's AI auditing agent found 24 Android bugs, and its author says a researcher should still check each one

    Article · · GitHub Security Blog · Kevin Stubbings · Sep 28, 2026

  8. A planted page note got browser agents to authorize access in all 73 attack runs; a server-side check stopped it

    Paper · · arXiv · Md Imran Hossen · Oct 10, 2026

  9. OWASP's agent Top 10 blames identity systems built for people, and does not say how it chose or ranked its risks

    Article · · OWASP GenAI Security Project · John Sotiropoulos et al. · Dec 9, 2025

  10. DIF's agent spec names every AI agent by a DID; the IETF's Agent Name Service switched to domain names

    Article · · Decentralized Identity Foundation (KYA-OS) · Dylan Hobbs et al. · Jul 29, 2026

  11. Coetzee's agent token is good for one declared call, and the hard part is hashing that call the same way twice

    Draft · · IETF Datatracker · Rudolf Jacobus Coetzee · Jul 19, 2026

  12. Atlassian's first draft for offline delegation banned bearer handles; revision 05 just recommends sender binding

    Draft · · IETF Datatracker · Larry Zhu et al. · May 22, 2026

  13. An open spec for stopping agent actions before they run needs the vendor's help to cover ChatGPT or Copilot

    Paper · · arXiv · Herman Errico · Feb 10, 2026

  14. A third-party agent talked a trusted one into unlocking the door; SEAgent blocks it, given the right labels

    Paper · · arXiv · Zimo Ji et al. · Jan 17, 2026

  15. Auth0's case for Token Vault: take the user ID out of the token call and let the session prove whose token it is

    Article · · Auth0 Blog · Abhishek Hingnikar, Okta · Nov 4, 2025

  16. MCP's June 2025 authorization spec: a server accepts only tokens issued for it and passes none upstream

    Article · · Model Context Protocol project · Jun 18, 2025

  17. A fake error message relayed between agents got three multi-agent frameworks to run an attacker's code

    Paper · · arXiv · Harold Triedman et al. · Mar 15, 2025

Related terms

What appears alongside Confused deputy in the same briefings.

Audience restrictionBearer tokenAgent delegation