[matt4077]

Apple and others have poured a lot of resources into a concept called “differential privacy” that can mathematically guarantee privacy for data in a machine learning context.

It’s rather well known, so you should probably read up on it before espousing strongly worded opinion weakly connected to the current state of the art.

[moultano]

I don't think anyone has managed to train differentially private models with acceptable performance, and releasing raw data in differentially private way seems impossible to me.