“Observation” only means that the AI has detected suspicious behavior; no punitive measures will be taken. So if you find that your risk-control status is “Observation,” there’s no need to appeal.