CVE-2026-62228: OpenClaw authz bypass

CISO Take

OpenClaw's node exec approval flow fails to validate that the gateway and the executing node share the same environment configuration, letting a lower-trust caller execute actions beyond what was actually approved for it. There's no EPSS score, no CISA KEV listing, and no public exploit or Nuclei template yet, so this is not under active exploitation today — but the CVSS 8.8 (network, low complexity, low privileges, no user interaction, full compromise of confidentiality, integrity and availability) means a modest foothold is enough to fully escalate within an agentic execution environment. OpenClaw is an AI agent orchestration package, so this directly threatens any deployment where a gateway routes exec approvals to distributed nodes on behalf of agents or users with tiered trust levels. Patch to OpenClaw 2026.6.5 or later immediately; until patched, audit that gateway and node environment configs are identical/pinned, and review approval logs for exec actions that crossed environment boundaries or exceeded the caller's declared scope.

Sources: NVD GitHub Advisory CISA KEV ATLAS

What is the risk?

High risk despite the absence of confirmed in-the-wild exploitation. The CVSS 3.1 vector (AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H) shows this is remotely reachable, trivially exploitable by anyone with low (not high) privileges, requires no user interaction, and yields full compromise of confidentiality, integrity, and availability once triggered. The lack of an EPSS percentile and KEV listing suggests it hasn't been weaponized publicly yet, and no PoC or scanner template exists — but authorization-bypass bugs in agent exec paths are attractive targets because they convert a low-trust identity directly into elevated command execution, which is a high-value primitive for supply-chain and lateral-movement attacks in agentic infrastructure.

How does the attack unfold?

Low-trust access acquired

Attacker obtains or controls a caller identity with limited, lower-tier approval scope on the OpenClaw gateway.

AML.T0012

Cross-environment exec request

Attacker submits a node exec request that targets a node whose environment configuration differs from the gateway's expected configuration.

AML.T0053

Authorization check bypassed

The mismatched gateway/node environment configs cause the authorization check to fail open, letting the request execute beyond its approved scope.

AML.T0107

Unauthorized execution and persistence

The attacker's actions run with elevated authorization on the node, enabling unauthorized command execution or persistence within the agent environment.

AML.T0050

Low-trust access acquired

Attacker obtains or controls a caller identity with limited, lower-tier approval scope on the OpenClaw gateway.

AML.T0012

Cross-environment exec request

Attacker submits a node exec request that targets a node whose environment configuration differs from the gateway's expected configuration.

AML.T0053

Authorization check bypassed

The mismatched gateway/node environment configs cause the authorization check to fail open, letting the request execute beyond its approved scope.

AML.T0107

Unauthorized execution and persistence

The attacker's actions run with elevated authorization on the node, enabling unauthorized command execution or persistence within the agent environment.

AML.T0050

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

0.5%

chance of exploitation in 30 days

Higher than 38% 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

Upgrade OpenClaw to version 2026.6.5 or later immediately. Until patched, audit gateway and node deployments to ensure environment configurations (trust tiers, scopes, approval policies) are consistent and pinned across all gateway/node pairs — do not mix staging/production or differently-scoped nodes behind the same gateway. Review exec approval logs for actions that originated from lower-trust callers but executed with elevated scope, or for approvals that crossed environment boundaries. Add detection for exec requests routed to nodes whose environment tag doesn't match the gateway's expected tag. Restrict which nodes a given gateway can dispatch to via network segmentation as defense-in-depth.

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 AML.T0050 - Command and Scripting Interpreter AML.T0053 - AI Agent Tool Invocation AML.T0107 - Exploitation for Defense Evasion AML.T0112.000 - Local AI Agent

Which compliance frameworks are affected?

This CVE is relevant to:

EU AI Act

Article 15 - Accuracy, robustness and cybersecurity

ISO 42001

A.6.2 - AI system access control

NIST AI RMF

MANAGE 2.3 - Procedures are followed to respond to and recover from AI system risks based on established access and authorization controls

OWASP LLM Top 10

LLM06:2025 - Excessive Agency

Frequently Asked Questions

What is CVE-2026-62228?

OpenClaw's node exec approval flow fails to validate that the gateway and the executing node share the same environment configuration, letting a lower-trust caller execute actions beyond what was actually approved for it. There's no EPSS score, no CISA KEV listing, and no public exploit or Nuclei template yet, so this is not under active exploitation today — but the CVSS 8.8 (network, low complexity, low privileges, no user interaction, full compromise of confidentiality, integrity and availability) means a modest foothold is enough to fully escalate within an agentic execution environment. OpenClaw is an AI agent orchestration package, so this directly threatens any deployment where a gateway routes exec approvals to distributed nodes on behalf of agents or users with tiered trust levels. Patch to OpenClaw 2026.6.5 or later immediately; until patched, audit that gateway and node environment configs are identical/pinned, and review approval logs for exec actions that crossed environment boundaries or exceeded the caller's declared scope.

Is CVE-2026-62228 actively exploited?

No confirmed active exploitation of CVE-2026-62228 has been reported, but organizations should still patch proactively.

How to fix CVE-2026-62228?

Upgrade OpenClaw to version 2026.6.5 or later immediately. Until patched, audit gateway and node deployments to ensure environment configurations (trust tiers, scopes, approval policies) are consistent and pinned across all gateway/node pairs — do not mix staging/production or differently-scoped nodes behind the same gateway. Review exec approval logs for actions that originated from lower-trust callers but executed with elevated scope, or for approvals that crossed environment boundaries. Add detection for exec requests routed to nodes whose environment tag doesn't match the gateway's expected tag. Restrict which nodes a given gateway can dispatch to via network segmentation as defense-in-depth.

What systems are affected by CVE-2026-62228?

This vulnerability affects the following AI/ML architecture patterns: agent frameworks, multi-node agent orchestration, tool/code execution pipelines.

What is the CVSS score for CVE-2026-62228?

CVE-2026-62228 has a CVSS v3.1 base score of 8.8 (HIGH). The EPSS exploitation probability is 0.45%.

What is the AI security impact?

Affected AI Architectures

agent frameworksmulti-node agent orchestrationtool/code execution pipelines

MITRE ATLAS Techniques

AML.T0050 Command and Scripting Interpreter

AML.T0053 AI Agent Tool Invocation

AML.T0107 Exploitation for Defense Evasion

AML.T0112.000 Local AI Agent

Compliance Controls Affected

EU AI Act: Article 15

ISO 42001: A.6.2

NIST AI RMF: MANAGE 2.3

OWASP LLM Top 10: LLM06:2025

What are the technical details?

Original Advisory

OpenClaw before 2026.6.5 contain an authorization bypass vulnerability in node exec approvals that allows lower-trust callers to execute actions beyond their intended authorization by using different gateway and node environments. Attackers can exploit mismatched environment configurations to persist or execute actions that exceed the caller's approved permissions.

Exploitation Scenario

An attacker who controls or compromises a low-trust caller (e.g., a restricted-scope agent, a customer-tier API identity, or a sandboxed sub-agent) submits node exec requests through the gateway. Because the gateway and the target node don't validate that they share the same environment configuration, the node accepts and executes the request as if it had come from a higher-trust context, granting the attacker actions — command execution, persistence, or resource access — well beyond what their original approval scope allowed.

Weaknesses (CWE)

CWE-863 Incorrect Authorization Primary CWE-863 Incorrect Authorization

CWE-863 — Incorrect Authorization: The product performs an authorization check when an actor attempts to access a resource or perform an action, but it does not correctly perform the check.

[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.