The easiest way is to always use a DNS whitelist firewall on a external device on your network. It is a pain in the ass to manage that list, but you'll log and block every unexpected connection.

Otherwise, you can do a basic grep for "http" in the source, or even by scanning a binary for strings. This vector is easily mitigated by an attacker using encryption or a simple cypher, but it will catch a lot of script kiddies.

I mean someone pointing out a vulnerability in a piece of software should be a falsifiable claim, e.g. "they store their passwords in plaintext"—if it's foss then just look at the source. You don't need to read the entire source because you have been given a specific part of the code to look at. You need to only look at the process between the software receiving a password and its query to the database.

And if it's not foss I don't use it, and the claim may be unfalsifiable for an outsider who isn't bothered to try reverse engineering.

You can just look up the information on the web, but how do you know whether to trust that information?

With open source, it's easy to get caught lying, or even just pushing on the boundaries of good ethics (cough Ubuntu cough).

If it's controversial, it is possible to verify the claims.

but how do you know whether to trust that information

test it, break it, see what happens.

we either put it in virustotal, spin up a VM, and run the thing without network access, and just hope for the best