I'm using the two phrases synonymously because the GDPR protects personal information, which is the superset (the distinction of PII as a subset seems to be mor

I'm using the two phrases synonymously because the GDPR protects personal information, which is the superset (the distinction of PII as a subset seems to be more relevant in the US). The GDPR is concerned with privacy. If you derive information from my personal information in a way that makes it impossible to go back from the derived information to identifying me as an individual, it's anonymous and thus not relevant to my privacy. However if you use this exact same process but only use it once and record that you only did it with my information, it becomes linked to me again and can no longer be considered anonymous. For another example: imagine you have a closed group of 1000 participants. "One of the guys with blonde hair" is probably a fairly ambiguous identifier because you'd expect there to be more than one person in the group that description could apply to. However if I'm the only blonde guy in that group, it's now clearly referring to me as an individual and thus affecting my privacy. The thing about storing IP plus timestamp on the other hand is that while it may be practically anonymous to you, you're storing it. Even though you can't resolve that information to a single person right now, someone else could if you gave them access to it. You can make an argument about where exactly the line should be drawn considering it's rarely impossible that someone somewhere could use seemingly innocuous data to identify someone but it's not much of a leap to go from an IP address and a time stamp to a subscriber who might be a single individual: IPs are publicly registered to ISPs and those ISPs know who they assigned the IP to at a given point in time (especially in countries like Germany where ISPs are required to keep records of this) so you can already easily convert "IP plus timestamp" to "IP plus timestamp plus an organisation that is capable of resolving that IP plus timestamp to a subscriber". In other words: at best an IP plus timestamp isn't anonymous, it's at least pseudonymous (even if you have no legal means of resolving that pseudonym). FWIW there are free privacy policy generators out there for small scale websites (e.g. most blogs). Here's a good one for Germany: https://datenschutz-generator.de/ https://datenschutz-generator.de/ EDIT: For programmers: You have a set (IP, timestamp). IPs are publicly registered to ISPs so you can resolve that to an ISP, which can act as a function (IP, timestamp) => subscriber. A subscriber can be, among other things, a single individual. So storing the IP and timestamp is the equivalent of storing an identifier from a lookup table of subscribers (some of whom are single individuals). Whether the result of the lookup table is accessible by legal means (e.g. a warrant) or technical means (e.g. a decryption key) or practical means (e.g. a literal key to a safe) makes no difference.