Share Facebook Twitter LinkedIn Pinterest Email New research indicates that current AI agents are still susceptible to prompt injection attacks, potentially compromising their security and integrity.