Critical LangSmith Account Takeover Vulnerability Puts Users at Risk
A critical vulnerability, identified as CVE-2026-25750, has been discovered in LangSmith by Miggo Security researchers. This vulnerability poses a risk of token theft and potential complete account takeover.
A critical vulnerability, identified as CVE-2026-25750, has been discovered in LangSmith by Miggo Security researchers. This vulnerability poses a risk of token theft and potential complete account takeover.
LangSmith, a platform for debugging and monitoring large language model data, processes billions of events daily, highlighting the significance of this security flaw.
The vulnerability originates from an insecure API configuration in LangSmith Studio. The platform allows developers to use a flexible baseUrl parameter to direct frontend applications to different backend APIs. Previously, the application did not validate the destination domain of this parameter.
This lack of validation allowed malicious entities to exploit the flaw. If a user accessed a compromised link, their browser could inadvertently route API requests and session credentials to a hostile server.
A critical vulnerability, identified as CVE-2026-25750, has been discovered in LangSmith by Miggo Security researchers.
The attack does not rely on traditional phishing methods. Instead, it silently executes in the background using the victim's active session. The process initiates when an authenticated user visits a malicious or compromised webpage. The script then loads a crafted LangSmith Studio URL, pointing to an attacker-controlled server.
The victim's browser sends active session credentials to the malicious domain, providing a five-minute window for the attacker to hijack the account before the token expires.
Control over a LangSmith account allows attackers to access detailed AI trace histories, potentially exposing proprietary source code, financial records, or sensitive customer information. Attackers can also alter project settings or delete critical workflows.
LangChain has addressed the vulnerability by implementing a strict allowed origins policy. Domains must be pre-configured as trusted origins in the account settings to be accepted as an API base URL. Unauthorized requests are blocked.
According to the LangSmith Security Advisory released on January 7, 2026, there is no evidence of active exploitation. Cloud customers require no action, as the issue was resolved on the LangSmith Cloud platform by December 15, 2025. Self-hosted administrators should upgrade to LangSmith version 0.12.71 or Helm chart langsmith-0.12.33 and later to secure their environments.
Based on reporting by Cyber Security News.