Nor does the typical OpenAI paper. Three retractions out of ~300 manuscripts doesn't mean multiple retractions per manuscript.