What is end-to-end encryption and what does it not protect?
End-to-end encryption means content is encrypted on the sender device and decrypted only on the recipient device. The service carrying it holds only ciphertext and cannot read the contents, even under legal compulsion, because it does not possess the keys.
This is a meaningful guarantee and a specific one. It protects the content of your messages from the provider, from anyone intercepting traffic, and from an attacker who breaches the provider servers.
What it does not protect is broader than people assume.
- Metadata. Who you contacted, when, how often and from roughly where is generally not encrypted, because the service needs it to deliver messages. This information is frequently more revealing than content.
- Endpoints. Encryption protects data in transit, not the devices at either end. Anything that can read your screen or your storage sees the decrypted messages. A compromised phone defeats it entirely.
- Backups. This is the most common practical gap. If messages are backed up to a cloud service without their own encryption, the backup may be readable even though the messages were not. Several major platforms make encrypted backup optional rather than default.
- Anyone in the conversation. The recipient can screenshot, forward or simply be untrustworthy.
It is also worth knowing that some services encrypt in transit and at rest but hold the keys themselves, which is meaningfully different and often described in language that sounds similar. If the provider can show you your message history on a new device after a password reset alone, they can read it.
Verification features exist in serious messengers for a reason: they let you confirm you are encrypting to the right person rather than to an intermediary.