Or suppose I'm a Russian dissident in Moscow with a list of Ukrainian-government and military contacts on my contact list. I open FairEmail and suddenly my phone does DNS lookups for a dozen sensitive Ukrainian domains. Not great.
Edge cases like this can be multiplied.
It's not malicious. But it's not great. And it seems reasonable to me for the Play Store to require a disclosure before an app does this.
So that favicon feature should be an optional feature but calling it sharing contact details is really exaggerating it.
EDIT: It is even optional, so yes really it's a non-issue.
If a user turns on a feature to request favicons for received mail, it therefore seems logical that the app has to retrieve favicons. Doing direct retrieval is likely less damaging to privacy than using a central favicon retrieval server.
Maybe caching the favicons on a server hosted by the app creator so that no matter what domain the app looks up, it looks like it's contacting FairEmail.
I'm honestly a bit annoyed that OP is the top comment on HN when it exaggerates the impact of this feature.
The problem here is the app was deceptively processing contact lists without user consent, and when asked to get user consent, the dev refused. If you're a transparency- or privacy-branded app, then you doubly need to disclose this information to the user.
Google didn't save anyone, by discouraging an exceptional developer with this kind of privacy policy https://github.com/M66B/FairEmail/blob/master/PRIVACY.md from contributing anything further on to Android app ecosystem, while keeping all the data stealing apps and ad supported apps, and Facebook SDK using apps, in play store.
Someone being in your contacts list does not mean you have ever had any contact with them.
So the site only learns that the IP has someone on that domain among its contacts. They probably already know more than that, because they've seen some emails being exchanged. Not 100% tight, but definitely less problematic than using Gravatar.
The traffic can be sniffed as the certificate validation looks disabled and it falls back to HTTP. Meaning ISP can take a peek.
But it's definitely not malicious in any way.