Basically, a good way to never trust "it's okay, the data is anonymized" again is simply knowing what the "Hemisphere Program" is.

As an aside, I'm very much convinced that Signal's primary objective is to gather phone numbers in order to facilitate the US government tracing social networks of people who are already of interest Their main focus isn't on what these people are discussing, they want to know who is talking to whom first and foremost. Signal's subpar user experience is a feature from this perspective. Due to its inconvenience for the average person, those with a strong need or desire to communicate sensitive information are more likely to utilize it.