A cybersecurity threat where attackers craft and input deceptive prompts or commands into AI systems, such as chatbots or voice assistants, to manipulate the system’s output or actions for harmful purposes. This can lead to unauthorized access, data leaks, or the spreading of misinformation.