CVE-2026-32243: Discourse XSS Vulnerability

CVE-2026-32243 Overview

CVE-2026-32243 is a stored Cross-Site Scripting (XSS) vulnerability affecting Discourse, the popular open-source discussion platform. An attacker with the ability to create shared AI conversations can inject arbitrary HTML and JavaScript via crafted conversation titles. This malicious payload executes in the browser of any user viewing the onebox preview, potentially enabling session hijacking or unauthorized actions performed on behalf of the victim.

Critical Impact

Attackers can execute arbitrary JavaScript in victim browsers, potentially stealing session tokens, performing unauthorized actions, or compromising user accounts through the shared AI conversation feature.

Affected Products

• Discourse versions 2026.1.0-latest to before 2026.1.3
• Discourse versions 2026.2.0-latest to before 2026.2.2
• Discourse versions 2026.3.0-latest to before 2026.3.0

Discovery Timeline

• 2026-03-31 - CVE-2026-32243 published to NVD
• 2026-04-01 - Last updated in NVD database

Technical Details for CVE-2026-32243

Vulnerability Analysis

This vulnerability is classified as CWE-79 (Improper Neutralization of Input During Web Page Generation), commonly known as Cross-Site Scripting (XSS). The flaw exists in Discourse's handling of shared AI conversation titles within the onebox preview functionality.

When a user creates a shared AI conversation, the conversation title is not properly sanitized before being rendered in onebox previews. This allows an attacker to embed malicious HTML or JavaScript code within the title field. When other users view the onebox preview containing the malicious conversation reference, the injected script executes within their browser context.

The stored nature of this XSS makes it particularly dangerous, as the malicious payload persists on the server and can affect multiple users over time without requiring continued attacker interaction.

Root Cause

The root cause stems from insufficient input validation and output encoding in Discourse's shared AI conversation title handling. The application fails to properly sanitize user-supplied conversation titles before rendering them in onebox previews, allowing script injection through the title parameter.

Attack Vector

The attack is network-based and requires low privileges—specifically, the ability to create shared AI conversations. No user interaction is required beyond viewing the crafted onebox preview. An attacker would:

1. Create a shared AI conversation with a maliciously crafted title containing JavaScript payload
2. Share or reference the conversation in a context where onebox previews are generated
3. When victims view the onebox preview, the injected JavaScript executes in their browser session

The injected script runs with the victim's session privileges, enabling actions such as session token theft, account manipulation, or further spreading of malicious content.

Detection Methods for CVE-2026-32243

Indicators of Compromise

• Unusual shared AI conversation titles containing HTML tags, <script> elements, or JavaScript event handlers
• Unexpected outbound requests to external domains originating from Discourse pages
• Reports of unusual account activity or unauthorized actions following conversation viewing

Detection Strategies

• Implement content security policies (CSP) that log or block inline script execution
• Monitor server logs for conversation titles containing suspicious patterns such as <script>, onerror=, onload=, or javascript:
• Deploy web application firewall (WAF) rules to detect XSS payloads in request parameters

Monitoring Recommendations

• Enable detailed logging for shared AI conversation creation and modification events
• Configure alerting for CSP violation reports that may indicate XSS attempts
• Monitor for anomalous session activity following user interactions with shared conversations

How to Mitigate CVE-2026-32243

Immediate Actions Required

• Upgrade Discourse immediately to patched versions 2026.1.3, 2026.2.2, or 2026.3.0
• Review existing shared AI conversations for suspicious or malicious titles
• Implement strict Content Security Policy headers to mitigate potential script execution
• Consider temporarily disabling the shared AI conversation feature until patching is complete

Patch Information

Discourse has released security patches addressing this vulnerability in versions 2026.1.3, 2026.2.2, and 2026.3.0. The fix is available in commit cac7d618a562ce934f8dbf73cdb70066a4806b4c. Organizations should update to the latest patched version immediately.

For detailed patch information, refer to the GitHub Security Advisory GHSA-pjc5-8x3w-rfwx and the patch commit.

Workarounds

• Restrict the ability to create shared AI conversations to trusted users only until patching is complete
• Implement server-side input sanitization for conversation titles as an additional defense layer
• Deploy Content Security Policy headers with script-src 'self' to prevent inline script execution
• Use a Web Application Firewall (WAF) to filter requests containing common XSS payloads

# Example CSP header configuration for nginx
add_header Content-Security-Policy "default-src 'self'; script-src 'self'; style-src 'self' 'unsafe-inline'; img-src 'self' data:; frame-ancestors 'none';" always;