The bug allows attacker-controlled model servers to inject code, steal session tokens, and, in some cases, escalate to remote code execution on enterprise AI backends.
Results that may be inaccessible to you are currently showing.
Hide inaccessible results
Feedback