CVE-2026-92544

A model name could alter a gateway usage-accounting database query

Kong · Kong AI Gateway LLM usage counters

First documented disclosure · Review checkpoint

Summary

Usage accounting incorporated a model identifier in a database statement without sufficient separation. The client could execute unintended SQL with the gateway database account’s permissions.

Severity and exploitation evidence

High · 8.7/10 · CVSS 4.0. Severity source.

Exploitation not established: The reviewed advisory does not report malicious exploitation.

Exploitation evidence source

Technical severity is not a measure of observed attacks or of risk to your deployment. Attempts, demonstrations, and successful compromise retain the qualifications in the source.

How the weakness works

Usage accounting incorporated a model identifier in a database statement without sufficient separation.

Conditions required

An affected gateway processed a client-controlled model name using its LLM usage counters.

Potential impact

The client could execute unintended SQL with the gateway database account’s permissions.

Affected versions and configurations

Kong AI Gateway 2.2.0. Kong also lists five Enterprise release lines in its CNA record.

Fix and mitigation status

Kong AI Gateway 2.2.1. Enterprise fixes are branch-specific.

Practical next steps

Upgrade AI Gateway to 2.2.1 or apply the corresponding Enterprise branch update.

Operating lesson

Treat model names as untrusted data throughout usage accounting.

Evidence and limits

The CVE record was published on 2026-11-10.

Sources