CVE-2026-62663: Banks generates meaningful LLM prompts using a simple template language. In versions prior to 2.4.4, all four media filters (image, audio, video, document) in banks accept untrusted user input as f...

Added to the CISA Known Exploited Vulnerabilities catalog on 30 Jul 2026. Vendor: AI/ML. Product: LLM. CVSS score: 7.5.

Banks generates meaningful LLM prompts using a simple template language. In versions prior to 2.4.4, all four media filters (image, audio, video, document) in banks accept untrusted user input as file paths via Path(value) and pass them directly to open(file_path, "rb") without any path sanitizat...

Required action: Review and patch if applicable to your AI infrastructure.