Comment by superb_dev

What’s to stop an attacker from using prompt injection against this firewall? I don’t understand how your AI is anymore secure than the AI it’s protecting