Code Execution Vulnerability in Open GenAI Stack for Meta AI Products
CVE-2026-77177
9.8CRITICAL
What is CVE-2026-77177?
The Open GenAI Stack, notably utilized in Meta's AI backend for applications like WhatsApp, is susceptible to a prompt injection flaw. This vulnerability enables attackers to execute arbitrary code through server-side expression evaluation using Jinja2 template syntax, as the system fails to sanitize user inputs adequately. This oversight poses significant risks, potentially allowing malicious actors to compromise system integrity and functionality.
