New ENCFORGE Ransomware Targets AI Model Files in Langflow RCE Attack
A new ransomware, ENCFORGE, is targeting AI model files and infrastructure, leveraging a vulnerability in Langflow (CVE-2025-3248) to gain remote code execution. The ransomware, developed by a threat actor linked to a previous JADEPUFFER campaign, encrypts model weights, vector indexes, and training datasets, and uses a sophisticated technique to break into containerized environments. The attacker bypassed a blocked delivery route by exploiting a Docker socket, demonstrating a significant escalation of privileges. The campaign is attributed to a single actor, with no confirmed victim count, and the ransomware relies solely on encrypted data for extortion.
Summary written automatically in our own words from the original article, which belongs to its publisher and remains the reference. It may contain errors. Sources & data
