Take a look at the attached, and think about not the words that it says, but the outcomes of infrastructures which would be needed in order for them to be implemented:

• all images which are shared on social media, will need to be checked against registers of “prohibited content” and matches will have to be “reported”, which would be a fine way to track leaked whistleblower content eg: slides by Edward Snowden
• text-analysis, yet the technology should not be able “understand the substance” of the issue; computers don’t “understand” anything and natural-language-processing is computationally expensive on the endpoint; BUT sensitive text like “XKeyScore“, “Weekend Protest” or the Prime Minister’s phone number can be detected & reported
• Interactions that are covered by professional secrecy will not be spied upon. Reassuring. So there will be a registered list of officially-approved journalists, doctors, lawyers, and politicians, and their communications will be (of course) safe from snooping? Of course it’s not like politicians ever become abusers?

This boilerplate stands on its head; any communications mechanism which is end-to-end encrypted must never leak even 1 bit of content to non-participants.

This framing enables much more than that.