We're still expected to simply trust that they're sufficiently anonymized
Actually Google fought in court to make sure that it didn't have to anonymize its logs, ever. So it's a safe bet to assume that your logs are not anonymized.
Actually Google fought in court to make sure that it didn't have to anonymize its logs, ever. So it's a safe bet to assume that your logs are not anonymized.
We anonymize this log data by removing part of the IP address (after 9 months) and cookie information (after 18 months).
But to make regulators happy, after 18 months, they can't spit out your raw IP addresses. Feel better now?
In the greater context of things, however, it is not anonymous: based on your previous access patterns, and your current (< 18 months) access patterns, they could link your anonymized logs with your unanonymized logs. Is that still anonymous then ?
This all reminds me of some research from Germany I heard about a few years ago: basically, if you have someone's ZIP code, age, gender and education, you can get a > 99% accuracy resolving who this person is in real life.