AI agents can plan gym workouts, book flights and make decisions on our behalf. But granting them greater autonomy may pose serious consequences when AI agents go rogue. With AI researchers calling for tighter restrictions, what safety guardrails should Singapore put in place - and are we already doing that?