CVE-2026-62207: OpenClaw auth bypass exposes admin

CISO Take

OpenClaw before version 2026.6.5 fails to properly enforce policy checks on configured input paths, letting a caller with only low privileges reach tools that should require admin-level authorization. There's no EPSS score, no CISA KEV listing, and no public exploit or Nuclei template yet, so this hasn't been weaponized in the wild — but the CVSS 8.8 score reflects a network-exploitable, low-complexity flaw needing no user interaction that yields full confidentiality, integrity, and availability impact once triggered. In an agentic AI deployment, admin-scoped tools typically control configuration, credentials, or destructive operations, so a successful bypass effectively hands an attacker the keys to the agent's most privileged capabilities. Patch to OpenClaw 2026.6.5 or later immediately, and in the interim audit which callers can reach admin tool paths and add compensating authorization checks at the gateway/proxy layer. Monitor logs for low-privilege identities invoking admin-tagged tool calls as a detection signal until patched.

Sources: NVD GitHub Advisory ATLAS

What is the risk?

High severity (CVSS 8.8) with network attack vector, low complexity, and no user interaction required makes this an attractive target once details circulate — the only mitigating factor is the low bar (PR:L) already required to reach the vulnerable path, meaning some existing access is a prerequisite. No EPSS data, no CISA KEV entry, no public exploit code, and no Nuclei scanner template exist as of publication, indicating exploitation has not yet been observed or automated. However, authentication/authorization bypass vulnerabilities in AI agent frameworks are historically fast-moving once a PoC surfaces, because the impact (full C/I/A compromise via admin tool invocation) is severe and the exploit logic (missing policy check) is typically straightforward to reverse-engineer from the patch diff. Treat this as high risk pending patch deployment, not low risk because it's currently unweaponized.

How does the attack unfold?

Initial Access

An attacker with existing low-privilege access sends a request toward a tool invocation path in the OpenClaw agent deployment.

AML.T0053

Authorization Bypass

Insufficient policy checks on the configured input path fail to reject the low-trust request meant for admin-scoped tools.

AML.T0107

Privilege Escalation

The attacker's request is processed as if authorized, granting access to admin-scoped agent tools.

AML.T0012

Impact

The attacker performs admin-level actions via the compromised tool access, achieving full confidentiality, integrity, and availability impact.

Initial Access

An attacker with existing low-privilege access sends a request toward a tool invocation path in the OpenClaw agent deployment.

AML.T0053

Authorization Bypass

Insufficient policy checks on the configured input path fail to reject the low-trust request meant for admin-scoped tools.

AML.T0107

Privilege Escalation

The attacker's request is processed as if authorized, granting access to admin-scoped agent tools.

AML.T0012

Impact

The attacker performs admin-level actions via the compromised tool access, achieving full confidentiality, integrity, and availability impact.

What systems are affected?

Package	Ecosystem	Vulnerable Range	Patched
OpenClaw	pip	—	No patch
3 dependents 37% patched ~3d to patch Full package profile →

Do you use OpenClaw? You're affected.

How severe is it?

CVSS 3.1

8.8 / 10

EPSS

0.6%

chance of exploitation in 30 days

Higher than 45% of all CVEs

Source: EPSS v3 — FIRST.org

Exploitation Status

No known exploitation

Sophistication

Moderate

What is the attack surface?

AV Network

AC Low

PR Low

UI None

S Unchanged

C High

I High

A High

What should I do?

1 step

1) Upgrade to OpenClaw 2026.6.5 or later immediately — this is the vendor-confirmed fix. 2) Until patched, audit all configured input paths and tool-to-privilege mappings referenced in the GHSA advisory (GHSA-cf2p-f286-mphf) to identify which admin-scoped tools are exposed to lower-trust callers. 3) Add or tighten compensating authorization checks at a reverse proxy or API gateway in front of OpenClaw to enforce privilege separation independently of the application's internal policy engine. 4) Enable/review audit logging for tool invocations and alert on any low-privilege identity or token successfully calling an admin-tagged tool. 5) Rotate credentials/tokens for any admin-scoped tool that may have been reachable prior to patching, in case of undetected prior abuse.

What does CISA's SSVC say?

Decision Track

Exploitation none

Automatable No

Technical Impact total

Source: CISA Vulnrichment (SSVC v2.0). Decision based on the CISA Coordinator decision tree.

How is it classified?

Auth Bypass Code Execution Agent Plugin AML.T0012 - Valid Accounts AML.T0053 - AI Agent Tool Invocation AML.T0107 - Exploitation for Defense Evasion

Which compliance frameworks are affected?

This CVE is relevant to:

EU AI Act

Article 15 - Accuracy, robustness and cybersecurity

ISO 42001

A.6.2.4 - AI system security controls

NIST AI RMF

MANAGE-2.1 - AI system risks and benefits are managed via mechanisms to sustain value and address negative impacts

OWASP LLM Top 10

LLM06 - Excessive Agency

Frequently Asked Questions

What is CVE-2026-62207?

OpenClaw before version 2026.6.5 fails to properly enforce policy checks on configured input paths, letting a caller with only low privileges reach tools that should require admin-level authorization. There's no EPSS score, no CISA KEV listing, and no public exploit or Nuclei template yet, so this hasn't been weaponized in the wild — but the CVSS 8.8 score reflects a network-exploitable, low-complexity flaw needing no user interaction that yields full confidentiality, integrity, and availability impact once triggered. In an agentic AI deployment, admin-scoped tools typically control configuration, credentials, or destructive operations, so a successful bypass effectively hands an attacker the keys to the agent's most privileged capabilities. Patch to OpenClaw 2026.6.5 or later immediately, and in the interim audit which callers can reach admin tool paths and add compensating authorization checks at the gateway/proxy layer. Monitor logs for low-privilege identities invoking admin-tagged tool calls as a detection signal until patched.

Is CVE-2026-62207 actively exploited?

No confirmed active exploitation of CVE-2026-62207 has been reported, but organizations should still patch proactively.

How to fix CVE-2026-62207?

1) Upgrade to OpenClaw 2026.6.5 or later immediately — this is the vendor-confirmed fix. 2) Until patched, audit all configured input paths and tool-to-privilege mappings referenced in the GHSA advisory (GHSA-cf2p-f286-mphf) to identify which admin-scoped tools are exposed to lower-trust callers. 3) Add or tighten compensating authorization checks at a reverse proxy or API gateway in front of OpenClaw to enforce privilege separation independently of the application's internal policy engine. 4) Enable/review audit logging for tool invocations and alert on any low-privilege identity or token successfully calling an admin-tagged tool. 5) Rotate credentials/tokens for any admin-scoped tool that may have been reachable prior to patching, in case of undetected prior abuse.

What systems are affected by CVE-2026-62207?

This vulnerability affects the following AI/ML architecture patterns: agent frameworks, multi-tenant agent deployments, tool orchestration layers.

What is the CVSS score for CVE-2026-62207?

CVE-2026-62207 has a CVSS v3.1 base score of 8.8 (HIGH). The EPSS exploitation probability is 0.57%.

What is the AI security impact?

Affected AI Architectures

agent frameworksmulti-tenant agent deploymentstool orchestration layers

MITRE ATLAS Techniques

AML.T0012 Valid Accounts

AML.T0053 AI Agent Tool Invocation

AML.T0107 Exploitation for Defense Evasion

Compliance Controls Affected

EU AI Act: Article 15

ISO 42001: A.6.2.4

NIST AI RMF: MANAGE-2.1

OWASP LLM Top 10: LLM06

What are the technical details?

Original Advisory

OpenClaw versions before 2026.6.5 contain an authentication bypass vulnerability that allows lower-trust callers to reach admin-scoped tools. Attackers can perform actions requiring stronger authorization by exploiting insufficient policy checks on configured input paths.

Exploitation Scenario

An attacker who has obtained low-privilege access to an OpenClaw-based agent deployment — for example, a standard user API key or a scoped service account — crafts a request targeting a tool invocation path that is normally gated to admin-level callers. Because the policy check on that configured input path is insufficient, the request is processed rather than rejected, and the attacker's low-trust identity is treated as authorized to invoke the admin-scoped tool. From there, the attacker can perform actions reserved for administrators — altering agent configuration, accessing or exfiltrating sensitive data the tool has access to, or triggering destructive operations — achieving the full confidentiality, integrity, and availability impact reflected in the CVSS score, without needing any user interaction or elevated attack complexity.

Weaknesses (CWE)

CWE-862 Missing Authorization Primary CWE-862 Missing Authorization

CWE-862 — Missing Authorization: The product does not perform an authorization check when an actor attempts to access a resource or perform an action.

[Architecture and Design] Divide the product into anonymous, normal, privileged, and administrative areas. Reduce the attack surface by carefully mapping roles with data and functionality. Use role-based access control (RBAC) [REF-229] to enforce the roles at the appropriate boundaries. Note that this approach may not protect against horizontal authorization, i.e., it will not protect a user from attacking others with the same role.
[Architecture and Design] Ensure that access control checks are performed related to the business logic. These checks may be different than the access control checks that are applied to more generic resources such as files, connections, processes, memory, and database records. For example, a database may restrict access for medical records to a specific database user, but each record might only be intended to be accessible to the patient and the patient's doctor [REF-7].

Source: MITRE CWE corpus.