Indirect prompt injection is used to make chatbots output confidential information or perform malicious actions. Security researcher Johan Rehberger has recently devised and published a method to hack ...