Surely that's only their interpretation of privacy laws, and not something tested in courts.
It seems unlikely to actually break regulations if it's clear that the data has been fished out of the entropy well.
It seems unlikely to actually break regulations if it's clear that the data has been fished out of the entropy well.
I don't think AI providers get to hide behind an "entropy well" defense when that entropy is a direct consequence of AI professionals' greed and laziness around data governance.