Vulnerability · Published Oct 1, 2025 · Assigned by GitHub_M

CVE-2025-58055Discourse AI Suggestions Contain Insecure Direct Object Reference

Medium 4.3

In plain English

A medium-severity improper access control in discourse from discourse. There are no confirmed reports of exploitation so far. The chance it is exploited in the next 30 days is estimated at 0.3% (higher than 18% of all known vulnerabilities).

Severity (CVSS 3.1)4.3Medium

Exploit odds (30 days)0.3%top 82.2% of all CVEs

Exploited in the wildNo reportsCISA SSVC: none

Attack toolingNoneno public tooling known

Official description

Discourse is an open-source community discussion platform. In versions 3.5.0 and below, the Discourse AI suggestion endpoints for topic “Title”, “Category”, and “Tags” allowed authenticated users to extract information about topics that they weren’t authorized to access. By modifying the “topic_id” value in API requests to the AI suggestion endpoints, users could target specific restricted topics. The AI model’s responses then disclosed information that the authenticated user couldn’t normally access. This issue is fixed in version 3.5.1. To workaround this issue, users can restrict group access to the AI helper feature through the "composer_ai_helper_allowed_groups" and "post_ai_helper_allowed_groups" site settings.

Affected products (as reported)

Vendor	Product	Versions
discourse	discourse	< 3.5.1

Weakness: CWE-284, CWE-639 · Vector: CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:L/I:N/A:N

In plain English

Official description

Affected products (as reported)

References