Skip to content
SYNTHESISCYBER

Architecture

What end-to-end encryption does not protect

Encryption is not a single property that a product either has or does not have. Knowing where it stops is what makes it useful.

Architecture
Synthesis Security Team8 min read

Encryption is often described as a switch. In practice it is a set of specific guarantees about specific data at specific moments, and the interesting part is always the boundary.

Three different things people mean

Encryption in transit protects data while it moves between your device and a server. Encryption at rest protects data stored on disks the provider controls. End-to-end encryption means the provider holds only ciphertext and does not hold the key.

The first two protect against network interception and physical disk theft. Only the third means the provider cannot read your content even if compelled to try.

What stays visible regardless

Even with end-to-end encryption, certain things are unavoidably known:

  • Metadata: that an account exists, roughly when it is active, how many items it holds, and how large they are.
  • Anything you deliberately share outside the encrypted boundary, including exports and screenshots.
  • Anything on a device that is already compromised. Encryption protects data in storage and transit, not a machine with malware watching the screen.

The cost of holding the key yourself

A genuine zero-knowledge design has a real trade-off: if you lose the key, nobody can recover the data. That is not a bug or a support failure, it is the direct consequence of the provider being unable to read your content.

Any service that offers both zero-knowledge encryption and full account recovery without a key or recovery kit is describing something else. Read that claim carefully whenever you see it.