Given the openness of their model, that should be pretty easy to detect. If it were even a small possibility, wouldn’t openAI be talking about it very very loudly?