Artificial intelligence is now present in almost all applications we interact with on a daily basis, making communication and interaction easier for us and our customers – but it also opens new doors for attackers. In the past, cybercriminals used social engineering to persuade employees to disclose valuable information or reset passwords. Today, this is achieved through prompt injection of AI systems, and the disclosure of information or the execution of activities is often accomplished. Orange Cyberdefense presents current research findings and shows you what options you have for dealing with this vulnerability.