CVE-2026-44779: Discourse Bot Debug Endpoints Expose Whisper Translation Logs

Technical summary

CVE-2026-44779 is an information disclosure vulnerability (CWE-200) affecting Discourse versions 2026.1.0 through 2026.1.3, 2026.3.0 through 2026.3.0, and 2026.4.0 through 2026.4.0. The vulnerability exists in bot debug endpoints that fail to properly restrict access to whisper translation audit logs. An authenticated attacker can invoke these debug endpoints to retrieve audit trail data for translated whispers, bypassing the intended access controls. The CVSS v3.1 base score of 4.3 (MEDIUM, network-accessible, low privileges required, confidentiality impact only) reflects the requirement for authentication and limited scope of disclosure. Patches are available in versions 2026.1.4, 2026.3.1, 2026.4.1, and 2026.5.0-latest.1.

Business impact

For organizations running Discourse as an internal or customer-facing community platform, this vulnerability poses a privacy and compliance risk. Whispers are frequently used for sensitive moderation notes, private administrative discussions, and confidential user feedback. Exposure of translation audit logs could reveal the content and context of these private conversations, potentially violating user privacy expectations and regulatory requirements (such as GDPR or HIPAA in certain deployments). The breach of confidentiality is especially concerning if whispers contain personally identifiable information, moderation decisions, or business-sensitive data. However, the attack requires authentication, limiting the threat to insider actors or users with legitimate platform access.

Affected systems

Discourse installations running versions 2026.1.0–2026.1.3, 2026.3.0–2026.3.0, or 2026.4.0–2026.4.0 are affected. Environments running later patched versions (2026.1.4+, 2026.3.1+, 2026.4.1+, or 2026.5.0-latest.1+) are not vulnerable. Self-hosted Discourse deployments are the primary concern, though Discourse-as-a-Service offerings may also be affected depending on their version management practices. Administrators should verify their current Discourse version against the affected ranges.

Exploitability

Exploitation requires valid authentication credentials on the Discourse instance and knowledge of or discovery of the vulnerable debug endpoints. The attack is straightforward once the endpoint is identified—no special tooling or complex interaction is needed. An authenticated user can directly query the endpoint to retrieve whisper translation logs. The practical risk is moderated by the authentication requirement, but any user with a legitimate account (including low-privileged users such as regular community members) can exploit this. The vulnerability does not appear in CISA's Known Exploited Vulnerabilities catalog, suggesting no widespread active exploitation at time of publication, though the simplicity of exploitation means security researchers and motivated actors could weaponize it quickly.

Remediation

Apply patches immediately: upgrade to Discourse version 2026.1.4 or later (if on the 2026.1 branch), 2026.3.1 or later (if on the 2026.3 branch), 2026.4.1 or later (if on the 2026.4 branch), or 2026.5.0-latest.1 (if on the latest branch). These versions disable or properly restrict access to the vulnerable debug endpoints. For organizations unable to patch immediately, restrict API access and debug endpoint availability to trusted administrative IPs and consider disabling bot-related debug features if feasible. Monitor access logs for unexpected queries to debug endpoints.

Patch guidance

Discourse maintains parallel release branches; determine which branch your installation follows (2026.1, 2026.3, 2026.4, or latest/2026.5) and upgrade to the corresponding patched version listed in the description. Most self-hosted deployments use standard Docker or package-manager-based updates. Test patches in a staging environment first, particularly if you have custom bot integrations. The patches do not require data migration or configuration changes. Verify post-upgrade that bot debug endpoints are no longer accessible to non-administrative users by testing API calls to those endpoints.

Detection guidance

Monitor Discourse application logs and API access logs for requests to bot debug endpoints, particularly those containing 'whisper' or 'translation' parameters. Implement alerting on repeated or unusual access to debug endpoints from non-administrative users. If you have webhook or audit logging enabled, review historical access to these endpoints for signs of prior unauthorized disclosure. Check the Discourse Admin panel under "Logs" for any API calls that accessed the vulnerable endpoints during the vulnerable version window. Correlate timestamps with potential insider activity or suspicious user accounts.

Why prioritize this

While the CVSS score is MEDIUM (4.3), the vulnerability should be prioritized based on organizational context. If your Discourse instance stores sensitive information in whispers (moderation notes, user PII, compliance-related data), patch immediately—this is a direct confidentiality breach of intended-private data. The authentication requirement and lack of active exploitation lower urgency for lower-risk deployments (e.g., public community forums with no sensitive whispers), but patching remains important. The simplicity of exploitation and the privacy-sensitive nature of the disclosed data warrant swift remediation across most deployments.

Risk score, explained

The CVSS v3.1 base score of 4.3 (MEDIUM) reflects: network accessibility (AV:N), low attack complexity (AC:L), low attack privileges required (PR:L), no user interaction (UI:N), unchanged scope (S:U), and low confidentiality impact (C:L) with no integrity or availability impact (I:N, A:N). The score appropriately captures that authentication is required and only confidentiality is affected. However, the true business risk may exceed the numeric score if whispers contain high-value sensitive data; organizations handling regulated or classified information in whispers should treat this as higher priority despite the MEDIUM rating.

Frequently asked questions