French CERT Warns of Security Risks with Autonomous AI Personal Assistants in 2026

Since the beginning of 2026, autonomous AI personal assistants such as OpenClaw and Claude Cowork, along with other open-source solutions, have seen significant adoption on workstations. These tools differ from traditional conversational assistants by operating with higher autonomy and integration into user environments. The French CERT (CERT-FR) reported potential security risks associated with these agentic AI products in an advisory dated April 13, 2026. No specific vulnerabilities (e.g., CVE IDs) or technical attack vectors were disclosed in the notice. The alert highlights concerns over unauthorized data access, privilege escalation, and unintended execution of malicious actions due to their deep system integration.