As someone deeply involved in AI research, I strongly believe that if AI Safety Institutes identify signs of losing control over AI systems, labs must immediately pause training. Ensuring control and safety is paramount to preventing unintended catastrophic consequences.