In her piece regarding the utility of the AI agents that increasingly can do things, including interact with other AI agents and commit one’s personal resources without human involvement—given an initial set of instructions—Nicole Nguyen recounted her experiences testing these agents. In the course of that reporting, she had this:

Meta will soon give Muse an encrypted mode that will keep the company from being able to see what the user is doing with the agent.

It’s naive and dangerous to trust Meta’s—or any other such provider’s—word on this. We never really know two things about the Metas of the AI Agent world: whether they’ve installed a backdoor into the “encrypted” app, and how strong the app’s encryption really is.

That last, too, is at least partially beyond the control of the Metas of this world: quantum computing is on the verge of blowing up encryption generally, and post-quantum encryption is not yet ready.

A bonus third thing: we don’t know how firmly these companies will resist the government’s warrantless/subpoena-less requests for personal or other private information.