Microsoft just put out a blog today on Securing AI models and the security controls they employ for training/learning etc. I know a lot...