The study shows that LLMs lead in information coverage but fall short in inferential restraint and safety characterization, and therefore AI interview tools require accompanying routine quality assessment mechanisms to monitor risky behaviors.