This module provides a framework for integrating AI providers and features into Drupal. The module doesn't sufficiently sanitize certain HTMX attributes, which... (CVE-2026-118552)

Package: drupal/ai
Fixed in: 1.4.11, 1.5.2
Published: Wed 28 Oct 2026, 17:10 Lisbon

In plain words

The Drupal AI module does not sufficiently sanitise certain HTMX attributes. This can allow an attacker to run malicious scripts in a user's browser through cross-site scripting (XSS).

Am I affected?

Look for drupal/ai and its version in composer.lock. Fixes are listed in 1.4.11 and 1.5.2, but no affected version range is supplied, so the advisory does not identify which other versions are vulnerable. No configuration, feature or database condition is named, so the package version is all there is to check.

What to do

Check which version you have, then update to a fixed release for your branch (1.4.11, 1.5.2):

composer show drupal/ai | grep versions
composer update drupal/ai

If your version constraint in composer.json keeps you below a fixed release, raise it first. Whether your configuration is affected, and the exact version ranges, are in the advisory itself.

The advisory's description

This module provides a framework for integrating AI providers and features into Drupal.

The module doesn't sufficiently sanitize certain HTMX attributes, which can lead to a cross-site scripting (XSS) vulnerability.

The advisory

From OSV.dev, which gathers the GitHub Advisory Database, FriendsOfPHP and Drupal security advisories; the summary is theirs. The description is OSV's record of the advisory. "In plain words", "Am I affected?" and "If you cannot update yet" are based on that advisory, and checked to name no version it does not.