When we collect a lot of personal data, say via the US Census, the goal is to glean important aggregate information and statistics, while somehow preserving the anonymity and privacy of the individual respondents. There’s a rigorous mathematical process for doing this – that’s actually not that hard to understand – called Differential Privacy. I’ll explain how it works.

In the news: iOS has a new location privacy setting; Google confirms it’s rolling out AI to Gmail; Windows makes it much harder to avoid creating a Microsoft Account; WhatsApp is rolling out AI in Europe with no way to opt out; Switzerland is considering undermining encrypted communications; 23andMe is going bankrupt – it’s time to delete your data; France rejects a backdoor mandate; and finally, I have a lot to say about the US officials’ Signal chat debacle.