Attackers Exploit MLflow SSRF Flaw to Steal Cloud Credentials and Secrets

Two critical vulnerabilities in open-source platforms are under active exploitation, according to independent reports from watchTowr and VulnCheck. The first, CVE-2026-64849 (CVSS 9.3), is an unauthenticated Server-Side Request Forgery (SSRF) flaw in MLflow, an open-source AI platform. Attackers can reach the MLflow Tracking Server and issue HTTP requests to arbitrary internal cloud metadata endpoints, extracting sensitive data such as cloud credentials and secrets. The vulnerability affects MLflow versions prior to 3.15.0.

The second, CVE-2026-25895 (CVSS 9.5), is a missing authentication and path traversal vulnerability in FUXA, a web-based SCADA/HMI software for OT and industrial automation. This flaw allows unauthenticated remote attackers to write arbitrary files to the server file system, leading to remote code execution. Both vulnerabilities are being actively scanned and exploited in the wild.

Organizations using affected versions should apply patches immediately, monitor for suspicious activity, and restrict access to management interfaces. The exploitation of these flaws highlights the growing risk to AI and OT environments.