- CVSS
- MEDIUM · 5.4 v3.1 CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:L/I:L/A:N
- Published
- 2026-05-08
- Weakness
- CWE-284
- Source
- nvd.nist.gov/vuln/detail/CVE-2026-41487
Description
Langfuse is an open source large language model engineering platform. From version 3.68.0 to before version 3.167.0, there is a role-based-access control flaw in the LLM connection update flow. An authenticated, low-privileged user of role “member” in a project could request the update of an existing LLM connection to an attacker-controlled baseUrl, causing Langfuse to reuse the stored provider secret and redirect the test request to an attacker-controlled endpoint. This could expose the plaintext provider LLM API key for that connection. The attack is only possible if a user is already part of a project and has “member” scoped access. This issue has been patched in version 3.167.0.
References
- https://github.com/langfuse/langfuse/commit/7527bb0d84bc0a3dc24a4b16d22ed2e46e6dddff
- https://github.com/langfuse/langfuse/commit/e12386f9d4368bbfff24a4ad7fd53641091605ff
- https://github.com/langfuse/langfuse/pull/13027
- https://github.com/langfuse/langfuse/pull/13055
- https://github.com/langfuse/langfuse/releases/tag/v3.167.0
- https://github.com/langfuse/langfuse/security/advisories/GHSA-2524-j966-gfgh
How GTK Cyber trains on this
AI security training at GTK Cyber covers the LLM and ML-pipeline vulnerability classes that vulnerabilities like CVE-2026-41487 fall into. Our hands-on courses are taught by Charles Givre and other practitioners who break and defend production AI systems.