Cybersecurity researchers have demonstrated a new prompt injection technique called PromptFix that tricks a generative artificial intelligence (GenAI) model into carrying out intended actions by ...