AI Science Tracker

GitHub Security Lab AI agent finds tool-permission bypass in Open WebUI

Open WebUI · AI involvement attributed to GitHub Security Lab AI agent

Finding and attribution

Open WebUI v0.6.43 chat completion accepts caller-supplied tool IDs/servers without checking the user’s permission; server-held credentials may then invoke restricted tools with server privileges.

Product: Open WebUI. AI system: GitHub Security Lab AI agent. Team / organization: GitHub Security Lab.

CVE identifiers: CVE-2026-45350.

AI and human contribution (source account): GHSL credits an AI agent it developed, reviewed by Man Yue Mo and Peter Stöckli. GHSL records the report on 2026-01-05 and links the vendor advisory.

Human involvement is only described where the source account establishes it; unknown involvement does not mean none. Vendor confirmation of a flaw does not confirm every claim about AI contribution.

Discovery context: The public advisory documents the vulnerability and disclosure/remediation status. It does not claim exploitation in the wild.

Remediation status: fix documented. The vendor advisory reports local tool access fixed in v0.7.0 and the MCP server path fixed in v0.8.6; users on v0.8.6 or later are not affected.

A source-documented fix does not establish installed patches or current safety. Source checking does not independently reproduce the finding or establish observed exploitation.

Disclosed 15 May 2026. AI attribution dated 15 May 2026. Sources checked 15 Nov 2026; these are separate dates, not a live installation check.

Recorded CVSS: High · highest recorded score 7.1. Documented technical effects: other.

Severity, consequences and exploitation evidence

Case grouping uses the highest recorded CVSS score, with its version-specific severity band. It is descriptive, not a combined or editorial risk index. All recorded assessments remain below; scores from different versions and sources may disagree. CVSS specifications define version-specific scoring and severity bands.

CVE-2026-45350: other.
Prerequisites / affected scope: CNA description: Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to 0.8.6, there is a vulnerability in chat completion API, which allows attackers to bypass tool restrictions, potentially enabling unauthorized actions or access. In the chat_completion API, the parameters tool_ids and tool_servers are supplied by the user. These parameters are used to create a tools_dict by the middleware. This is then used by get_tool_by_id to retrieve the appropriate tool. However, there is no checks in that ensures the user that uses the API has permission to use the tool, meaning that a user can invoke any server tool by supplying the correct tool_id or tool_servers parameters via the chat completion API. Moreover, the authentication token stored in the server would be used when invoking the tool, so the tool will be invoked with the server privilege. This vulnerability is fixed in 0.8.6..
Exploitation evidence: Exploitation in the wild is not established by the cited description.. Impact source · checked 16 Nov 2026.

Documented fix: The vendor advisory reports local tool access fixed in v0.7.0 and the MCP server path fixed in v0.8.6; users on v0.8.6 or later are not affected.

Original evidence

Download this source record (JSON)

Open this case in the interactive tracker · Browse all static security cases