There is a retention policy for every bit of data we store and no data is kept longer than absolutely required! This is also true for backups!
We also anonymize and aggregate data whenever possible!
Some people always bring up that the data still exist in abstract form in ML models...but afaik most/all models were constantly rebuild with fresh data. I can't think of a model that still uses data from let's say 2 years ago.
Sensitive data such as location for example was only kept in encrypted form for a few minutes until the aggregation jobs had processed it. Such data stores were guarded like fort knox with multiple lines of defense!
There is a retention policy for every bit of data we store and no data is kept longer than absolutely required! This is also true for backups!
We also anonymize and aggregate data whenever possible!
Some people always bring up that the data still exist in abstract form in ML models...but afaik most/all models were constantly rebuild with fresh data. I can't think of a model that still uses data from let's say 2 years ago.
Sensitive data such as location for example was only kept in encrypted form for a few minutes until the aggregation jobs had processed it. Such data stores were guarded like fort knox with multiple lines of defense!