Improper input sanitization in the framework can be exploited through the Shell tool, allowing attackers to modify system files and steal data.

The post Vulnerability in MS-Agent AI Framework Can Allow Full System Compromise appeared first on SecurityWeek.