CVE-2026-62223: OpenClaw authorization bypass

CISO Take

OpenClaw, an open-source AI agent framework, has an authorization bypass in its device-pair approval flow that lets a caller with only low-level privileges trigger actions reserved for higher-trust callers, with no user interaction required and a network-reachable attack surface (CVSS 8.8, CWE-863 Incorrect Authorization). Because device pairing is how OpenClaw grants a new client or endpoint trust to control the agent, a successful bypass can let an attacker persist unauthorized access or actions inside an agentic workflow that may already have tool-invocation and file/system privileges. There is no evidence of active exploitation (not in CISA KEV, no public PoC or Nuclei template, EPSS unavailable), but the vulnerability is network-exploitable with low complexity and no user interaction, which makes it an attractive target once weaponized. Patch to OpenClaw 2026.5.18 or later immediately on any instance where the device-pair approval feature is enabled and network-reachable; where patching must wait, disable or firewall the device-pairing endpoint and audit `admin_jobs`/session logs for approval events from unrecognized low-trust callers.

Sources: NVD GitHub Advisory ATLAS

What is the risk?

High risk despite the absence of confirmed exploitation. The CVSS 8.8 vector (AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H) reflects a low-complexity, no-user-interaction, network attack that yields full confidentiality, integrity, and availability impact — the ceiling is capped only by the requirement for low (not zero) privileges. Given OpenClaw is an AI agent framework, exploitation doesn't just leak data — it can grant an attacker legitimate-looking authority to invoke the agent's own tools and actions, which is a materially worse outcome than a typical web-app auth bypass. Absence from CISA KEV and no public exploit/Nuclei coverage lowers near-term opportunistic risk, but the vendor advisory + VulnCheck writeup indicate the flaw is well-understood, so a working exploit could surface quickly.

How does the attack unfold?

Initial Access

A low-privilege caller (existing low-trust account, device, or API key) reaches the network-exposed device-pair approval endpoint.

Exploitation

The attacker exploits the misconfigured input path in the approval feature to bypass the authorization check with low complexity and no user interaction.

AML.T0107

Unauthorized Action Execution

With elevated standing granted by the bypass, the attacker invokes or persists agent actions beyond their intended authorization scope.

AML.T0053

Impact

The attacker achieves full confidentiality, integrity, and availability impact by controlling or manipulating agent behavior and data via the improperly authorized access.

AML.T0108

Initial Access

A low-privilege caller (existing low-trust account, device, or API key) reaches the network-exposed device-pair approval endpoint.

Exploitation

The attacker exploits the misconfigured input path in the approval feature to bypass the authorization check with low complexity and no user interaction.

AML.T0107

Unauthorized Action Execution

With elevated standing granted by the bypass, the attacker invokes or persists agent actions beyond their intended authorization scope.

AML.T0053

Impact

The attacker achieves full confidentiality, integrity, and availability impact by controlling or manipulating agent behavior and data via the improperly authorized access.

AML.T0108

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

1) Upgrade to OpenClaw 2026.5.18 or later immediately — this is the primary and only complete fix. 2) If immediate patching isn't possible, disable the device-pair approval feature or restrict it to a trusted, non-routable network segment (firewall/VPN-only access) until patched. 3) Audit recent device-pairing approvals and session/action logs for approvals or executed actions originating from unexpected accounts or IPs, especially any low-privilege identity performing actions outside its normal scope. 4) Rotate any credentials or tokens paired/approved during the vulnerable window if unauthorized pairing is suspected. 5) Add detection for anomalous authorization patterns (low-priv identity → high-priv action) in agent action logs going forward, since CWE-863 (incorrect authorization) issues tend to recur in agent-approval workflows.

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.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

Annex A.6.2.2 - AI system access control and authorization

NIST AI RMF

MEASURE 2.7 - AI system security and resilience is evaluated and documented

OWASP LLM Top 10

LLM06:2025 - Excessive Agency

Frequently Asked Questions

What is CVE-2026-62223?

OpenClaw, an open-source AI agent framework, has an authorization bypass in its device-pair approval flow that lets a caller with only low-level privileges trigger actions reserved for higher-trust callers, with no user interaction required and a network-reachable attack surface (CVSS 8.8, CWE-863 Incorrect Authorization). Because device pairing is how OpenClaw grants a new client or endpoint trust to control the agent, a successful bypass can let an attacker persist unauthorized access or actions inside an agentic workflow that may already have tool-invocation and file/system privileges. There is no evidence of active exploitation (not in CISA KEV, no public PoC or Nuclei template, EPSS unavailable), but the vulnerability is network-exploitable with low complexity and no user interaction, which makes it an attractive target once weaponized. Patch to OpenClaw 2026.5.18 or later immediately on any instance where the device-pair approval feature is enabled and network-reachable; where patching must wait, disable or firewall the device-pairing endpoint and audit `admin_jobs`/session logs for approval events from unrecognized low-trust callers.

Is CVE-2026-62223 actively exploited?

No confirmed active exploitation of CVE-2026-62223 has been reported, but organizations should still patch proactively.

How to fix CVE-2026-62223?

1) Upgrade to OpenClaw 2026.5.18 or later immediately — this is the primary and only complete fix. 2) If immediate patching isn't possible, disable the device-pair approval feature or restrict it to a trusted, non-routable network segment (firewall/VPN-only access) until patched. 3) Audit recent device-pairing approvals and session/action logs for approvals or executed actions originating from unexpected accounts or IPs, especially any low-privilege identity performing actions outside its normal scope. 4) Rotate any credentials or tokens paired/approved during the vulnerable window if unauthorized pairing is suspected. 5) Add detection for anomalous authorization patterns (low-priv identity → high-priv action) in agent action logs going forward, since CWE-863 (incorrect authorization) issues tend to recur in agent-approval workflows.

What systems are affected by CVE-2026-62223?

This vulnerability affects the following AI/ML architecture patterns: agent frameworks, device/edge agent deployments.

What is the CVSS score for CVE-2026-62223?

CVE-2026-62223 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 frameworksdevice/edge agent deployments

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: Annex A.6.2.2

NIST AI RMF: MEASURE 2.7

OWASP LLM Top 10: LLM06:2025

What are the technical details?

Original Advisory

OpenClaw before 2026.5.18 contain an authorization bypass vulnerability in the device-pair approval feature that allows lower-trust callers to execute actions beyond their intended authorization. Attackers can exploit misconfigured input paths to execute or persist unauthorized actions when the affected feature is enabled and reachable.

Exploitation Scenario

An attacker who already holds a low-privilege account or a low-trust device connection to an OpenClaw instance (e.g., a guest integration, a partially-onboarded client, or a compromised low-privilege API key) targets the device-pair approval endpoint over the network. By manipulating the input path for the pairing/approval request — exploiting the misconfigured authorization check — the attacker gets the system to treat their request as if it came from a higher-trust caller, without needing any user click-through or additional credentials. Once the bypass succeeds, the attacker's device or session is granted elevated standing within the agent's trust model, letting them invoke or persist actions (tool calls, configuration changes, data access) that should have required explicit high-trust approval — effectively hijacking part of the agent's authorized action surface for command-and-control or data exfiltration.

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.