Overview
CVE-2025-65107 describes a medium severity security vulnerability affecting Langfuse, an open-source large language model (LLM) engineering platform. This vulnerability, if exploited, could allow for unauthorized account takeover under specific configurations. The vulnerability exists in versions 2.95.0 to before 2.95.12 and 3.17.0 to before 3.131.0. A successful attack requires an authenticated user to be tricked into clicking a specially crafted URL. Patches are available in versions 2.95.12 and 3.131.0.
Technical Details
The vulnerability stems from insecure SSO provider configurations. Specifically, when an explicit AUTH_<PROVIDER>_CHECK setting is not defined, the system is susceptible to Cross-Site Request Forgery (CSRF) or phishing attacks. An attacker could craft a malicious URL that, when visited by an authenticated Langfuse user, could potentially lead to account compromise. The absence of the AUTH_<PROVIDER>_CHECK parameter means there is insufficient validation of the origin of the request, making the system vulnerable.
CVSS Analysis
The Common Vulnerability Scoring System (CVSS) score for CVE-2025-65107 is 6.5, which is categorized as MEDIUM severity. This score reflects the potential impact of a successful account takeover, balanced against the need for user interaction (clicking a malicious link).
Possible Impact
A successful exploitation of this vulnerability could have serious consequences, including:
- Account Takeover: An attacker could gain complete control over a Langfuse user’s account.
- Data Breach: Sensitive data associated with the compromised account could be exposed or stolen.
- Malicious Activity: An attacker could use the compromised account to perform unauthorized actions within the Langfuse platform, potentially affecting LLM models or data pipelines.
- Reputational Damage: If exploited, the security incident could negatively impact the reputation of organizations using vulnerable Langfuse versions.
Mitigation and Patch Steps
The recommended mitigation steps are as follows:
- Upgrade Langfuse: Upgrade to version 2.95.12 or 3.131.0 (or later) immediately. These versions contain the necessary patches to address the vulnerability.
- Workaround (if immediate upgrade is not possible): Explicitly set the
AUTH_<PROVIDER>_CHECKconfiguration setting for your SSO provider. Consult the Langfuse documentation for specific details on how to configure this setting. - User Awareness: Educate users about the risks of phishing attacks and the importance of verifying the authenticity of links before clicking them.
