End-to-end encryption, client-side encryption, and server-side encryption are often grouped together. All three protect data, but they operate at different points and create different trust models.
Server-side encryption
With server-side encryption, data reaches the server in a form the service can process and is then encrypted for storage. This can protect disks or backups, but it does not automatically prevent the service operator from reading the content.
Client-side encryption
With client-side encryption, content is encrypted on your device before upload. The server receives encrypted data. Whether the operator can still decrypt it depends on how keys are generated and managed.
End-to-end encryption
End-to-end encryption describes a model in which only the intended endpoints can decrypt the content. In a messenger, for example, those endpoints are the sender and recipient, while the relay server should not be able to read the message.
Why client-side does not automatically mean E2EE
An application can encrypt data on the client and then transmit the key to the service operator as well. Encryption happened client-side, but the operator still has the ability to decrypt.
Transport encryption is a fourth layer
TLS protects data in transit. That is important, but it does not replace client-side or end-to-end encryption. A service can use TLS, server-side encryption, and E2EE simultaneously for different threats.
Comparing the trust models
- Server-side: you trust the service with readable content.
- Client-side: content is encrypted before upload; the key model determines operator access.
- End-to-end: only intended endpoints should be able to decrypt.
Metadata is a separate issue
Even strong end-to-end encryption does not automatically hide every piece of metadata. Timestamps, account identifiers, recipients, file sizes, or device information may remain visible depending on the system.
Recovery can change the model
If a provider can fully recover your decryption key, the mechanism should be explained clearly. Convenient recovery may be useful, but it can change the trust assumptions.
Priviot
For Priviot Sync, precise technical terminology matters more than the strongest-sounding label. Where the design provides client-side encryption, that should be stated accurately; end-to-end or zero-access claims should only be used if the architecture actually supports them. See Priviot Sync and Security & Transparency.
Conclusion
Different encryption models protect against different risks. Always ask three questions: Where is the data encrypted? Who holds the keys? Who can actually read the content during normal operation or recovery?