Specific procedures should be put in place for market surveillance authorities to ensure a harmonised approach to enforcement actions for AI systems that present a risk to health, safety and fundamental rights.