MLflow - Authentication Bypass (CVE-2026-2652)

CVE-2026-2652 is an authentication bypass vulnerability in MLflow versions 3.9.0 and earlier, where certain FastAPI routes remain unprotected even when authentication is enabled. This flaw allows unauthenticated remote attackers to submit jobs, read job results, cancel running jobs, and inject arbitrary trace data, potentially leading to unauthorized access and manipulation of experiments. The vulnerability stems from a mismatch between Flask and FastAPI authentication mechanisms, and is resolved in version 3.10.0. Attackers could exploit this issue to compromise the integrity and confidentiality of machine learning workflows managed by affected MLflow deployments.

CrowdSec has been tracking this vulnerability and its exploits since 24th of June 2026.

Insights from the CrowdSec network reveal that the attackers trying to exploit CVE-2026-2652 are composed of a fairly even mix of opportunistic and targeted actors. Some attackers employ preliminary reconnaissance, while others use indiscriminate scanning. Data from the CrowdSec community also indicates a gradual decrease in attacks targeting CVE-2026-2652. While still present in the wild, exploitation levels have dropped noticeably week-over-week. This may signal that the vulnerability is becoming less relevant or that defenses are improving fast enough for attackers to lose interest.

Attackers exploit this vulnerability by sending unauthenticated requests to FastAPI endpoints such as /ajax-api/3.0/jobs/ and /v1/traces, bypassing authentication controls and gaining unauthorized access to job management and trace ingestion APIs.