Comment by inkysigma

While there isn't a way to differentiate between scraping for training data and content accessed in response to a user request, I think you can block Googlebot-extended to block training access.