If all the available platforms allow for data sharing and tracking on judicial orders, then a perpetrator does not have a safe way to give threats, without the risk of being identified.
In my opinion, this is a deterrent in itself, the fear of getting caught.
Yes, personally even I would feel disturbed and anxious when somebody threatens me, but it would be too broad to classify mental disturbance as a safety issue. I am saying this having survived an episode where my father was threatened for life.
I agree that it's a half assed patched, but if a platform isn't responding to litigation or law enforcement requests, there is always the chance of bad actors weaponizing that loophole.
And it's not like Proton AG hasn't been linked to terror attacks. The perpetrator of the Bataclan Massacre used Protonmail to communicate with handlers, leading to the French government to require email platforms like Proton AG to honor French law enforcement requests [1].
[0] - https://www.wired.com/2015/11/isis-opsec-encryption-manuals-...
[1] - https://www.ejn-crimjust.europa.eu/ejnupload/DynamicPages/Fr...
Say what now?!