# CVE-2026-100589: OpenClaw sandbox bypass in browser tool

Source: https://junglewise.ai/threats/cve-2026-100589-openclaw-versions-before-2026-7-1-contain-a-sandbox-bypass  
Publisher: Junglewise Threat Intelligence, https://junglewise.ai/threats  
Published: 2026-09-26. Updated: 2026-09-26  
License: CC BY 4.0 with a link back

CVE-2026-100589 · Severity: high · CVSS 8.3 · Published 2026-09-26

OpenClaw is a tool that allows agents to interact with browsers and other systems. Versions before 2026.7.1 contain a sandbox bypass flaw in the browser tool that allows sandboxed sessions to access browser operations on paired nodes, even when host control is explicitly disabled. An attacker who can control a sandboxed agent's input can inspect or manipulate the connected browser profile, potentially accessing sensitive authenticated sessions and data.

Technologies: [Openclaw](https://junglewise.ai/threats/technologies/openclaw). Vendors: [Openclaw](https://junglewise.ai/threats/vendors/openclaw).

## Quick answers

**Is CVE-2026-100589 exploited in the wild?** No. Junglewise Threat Intelligence has no report of CVE-2026-100589 being exploited in the wild.

**How severe is CVE-2026-100589?** High severity, with a CVSS score of 8.3, published 26 September 2026.

**Which products and versions are affected?** Openclaw before 2026.7.1.

**Is there a fix for CVE-2026-100589?** A patch was published on 2026-07-01: Fixed in version 2026.7.1 Junglewise finds every place you run it, drafts the upgrade and applies it once you approve.

**How is CVE-2026-100589 attacked?** It is attacked over the network.

**Am I affected by CVE-2026-100589?** Check whether you run Openclaw at an affected version. Junglewise does that check for you around the clock: connect your code, cloud and domains, and it matches new vulnerabilities like this one against what you actually run and tells you when one reaches you.

## Technical details

The vulnerability is an authorization bypass (CWE-863) in the browser tool component where sandboxed sessions can reach host browser actions despite allowHostControl=false configuration. The attack requires low privileges (the ability to steer sandboxed agent input) and network access, with no user interaction needed. A successful exploit allows the attacker to access browser operations and authenticated state on paired nodes, bypassing the sandbox boundary intentionally set by the operator.

## Affected products

- OpenClaw OpenClaw before 2026.7.1

## Timeline

- 2026-09-11: disclosed: Security advisory published on GitHub
- 2026-07-01: patched: Fixed in version 2026.7.1

## References

- [https://github.com/openclaw/openclaw/security/advisories/GHSA-9x88-f7rh-4c83](https://github.com/openclaw/openclaw/security/advisories/GHSA-9x88-f7rh-4c83)
- [https://www.vulncheck.com/advisories/openclaw-before-2026.7.1-sandbox-bypass-via-browser-node](https://www.vulncheck.com/advisories/openclaw-before-2026.7.1-sandbox-bypass-via-browser-node)

## Related threats

- [OpenClaw incorrect authorization in mcp.app.view method](https://junglewise.ai/threats/cve-2026-102807-openclaw-before-2026-9-4-contains-an-incorrect-authorization)
- [OpenClaw incorrect authorization in Gateway media root allowlist](https://junglewise.ai/threats/cve-2026-102806-openclaw-before-2026-9-5-contains-an-incorrect-authorization)
- [OpenClaw exec approval bypass in Google Meet node commands](https://junglewise.ai/threats/cve-2026-100599-openclaw-versions-2026-5-1-through-2026-7-0-fail-to-apply-the)
- [OpenClaw Signal approval reaction binding logic error](https://junglewise.ai/threats/cve-2026-100598-openclaw-npm-package-openclaw-before-2026-7-1-incorrectly-binds)
- [OpenClaw time-of-check time-of-use race condition in filesystem operations](https://junglewise.ai/threats/cve-2026-100597-openclaw-npm-package-openclaw-before-2026-7-1-is-vulnerable-to-a)