Added "This includes training data - ML models should not be used if their training data has violated privacy." to privacy guidance.
This is in response to Christine Runnegar's input in the brainstorm: "one that may not be front of mind is what approach should web developers take regarding the training data, e.g. would the guidance say that Web ML should not use ML models based on training data that abused privacy?"
Not sure the wording as I've put it is the best - happy for anyone to suggest better.
Added "This includes training data - ML models should not be used if their training data has violated privacy." to privacy guidance.
This is in response to Christine Runnegar's input in the brainstorm: "one that may not be front of mind is what approach should web developers take regarding the training data, e.g. would the guidance say that Web ML should not use ML models based on training data that abused privacy?"
Not sure the wording as I've put it is the best - happy for anyone to suggest better.