Share Facebook Twitter LinkedIn Pinterest Email Researchers are advocating for AI agents to be considered untrusted systems due to inherent risks and potential vulnerabilities that require careful management and oversight.