Let me try a #nationalsecurity argument against #ChatControl. Tell me how I am doing. 

Compared to old telephone system, packet routed data networks are notoriously insecure. Routers are routinely hacked by not only nation states, but by “hobbyists” as well. This unsafe architecture makes end to end encryption not only desired, but necessary. Data leaks occur if they can occur, among others because of unworkable IT policies or government IT infrastructure not functioning properly. Humans are focused on their tasks which is not computer security. Hence, E2E encryption for the general public is a vital fallback in a multi layer defence against data leaks.

The only place to break properly implemented E2E is on the devices, before encryption or after decryption, i.e. on the device. 

So, let’s look at the centralised collection of messages, pictures etc, detected with a lot of false positives, by the magical non-existing on-device AI that will not drain batteries in any way, this will be a high value target for criminal hackers, state sponsored hackers, etc. So instead of forcing the bad guys to spend a lot of resources on hacking individuals, they can focus on the single, very likely badly protected goldmine of intelligence which the central collection place will be.