Critical Vulnerabilities in MLflow and FUXA Under Active Exploitation
Two critical vulnerabilities affecting **MLflow**, an open-source AI platform, and **FUXA**, an open-source SCADA/HMI software, are currently facing active malicious scanning and exploitation. These flaws, identified as **CVE-2026-64849** and **CVE-2026-25895**, pose significant risks, including data exfiltration and remote code execution, prompting urgent patching recommendations for affected organizations.
Independent reports from cybersecurity firms **watchTowr** and **VulnCheck** highlight ongoing malicious activities targeting two severe vulnerabilities in widely used open-source platforms.
## MLflow's SSRF Flaw (**CVE-2026-64849**)
The first vulnerability, **CVE-2026-64849**, impacts **MLflow** versions prior to 3.15.0. This unauthenticated Server-Side Request Forgery (SSRF) vulnerability carries a CVSS score of 9.3.
According to **watchTowr**, attackers are actively exploiting this flaw to reach internal cloud metadata services, enabling the exfiltration of sensitive cloud credentials and secrets. **watchTowr** observed widespread scanning for exposed **MLflow** instances online within hours of the CVE's assignment on August 17, 2026.
**Yordan Ganchev**, Principal Threat Intelligence Specialist at **watchTowr**, explained that the bug allows attackers to proxy requests through affected systems by exploiting a flaw in **MLflow**'s model-registry webhooks. This bypasses previous fixes due to how web redirects are handled, with global honeypot telemetry confirming attempts to extract credentials from cloud-hosted **MLflow** systems.
Organizations running **MLflow** are strongly advised to prioritize patching affected systems, review audit logs for any signs of compromise, and verify if sensitive credentials have been exposed.
## FUXA's Authentication Bypass and Path Traversal (**CVE-2026-25895**)
The second critical vulnerability, **CVE-2026-25895**, affects **FUXA** versions up to 1.2.9. With a CVSS score of 9.5, this flaw combines missing authentication for a critical function with a path traversal vulnerability.
**VulnCheck** detected malicious scanning for this vulnerability beginning August 18, 2026. A single IP address has been observed broadly scanning the internet for the approximately 60 publicly exposed **FUXA** installations.
**Caitlin Condon**, Vice President of Research at **VulnCheck**, noted that attacker requests attempt to overwrite `main.js` with junk data via the path traversal, though no Remote Code Execution (RCE) payloads have been observed yet.
This is not the first time **FUXA** has been a target. Over the past year, **CVE-2026-25939** and **CVE-2023-33831** have also seen active exploitation, with activity for the latter dating back to November 2025.