Last updated 19 October 2026. This is an English translation of the Finnish original; if the versions differ, the Finnish version applies.
Espero's job is to translate messages so that people who speak different languages understand each other. We describe our protections plainly and do not promise protection the service does not provide.
End-to-end encryption would stop the server from reading messages. Server-side translation requires processing the text. Ordinary Espero chats are built on translation, so we do not claim they are end-to-end encrypted.
Cloud translation sends message text to third-party AI services. Our providers are listed by recipient category in section 4 of the Privacy Policy, with processing countries and transfer mechanisms; a named list is available on request from hello@espero.live. Voice, video, image or interpreter input is sent only when that feature is used. Speech recognition and reading translations aloud currently take place in the United States. Translated call captions are not available.
Eligible cloud messages may receive semantic-search embeddings while the search feature is enabled. Summaries, AI answers, the assistant, pronunciation and transliteration requested by an authorised user process the relevant content. Searches and Ask Espero questions are processed when they are submitted.
We do not use your messages to train AI models; if we want to do so in the future, we will first ask for your separate consent. We do not sell content or target advertising with it. We use the paid APIs of AI services whose terms state that the providers do not use your content to train their own models; they may keep requests for a limited period for abuse monitoring.
| Capability | Espero |
|---|---|
| App/API transport encryption | HTTPS/TLS |
| Call-media transport encryption | WebRTC encryption |
| Storage encryption | Provider states encryption at rest |
| Server can read cloud messages for features | Yes |
| Espero sells content or uses it for advertising | No |
| Espero uses messages to train AI models | No |
One-to-one WebRTC calls travel encrypted between the devices or through a TURN relay. Our servers do not receive or store call audio or video. Calls are not recorded. The server handles the routing, participant, type, timing and state data needed to connect calls, keep call history and prevent abuse. Setting up a call reveals limited network information to the other participant and to the relay service used.
So call media is not readable by the server the way cloud message text is. We still do not formally call calls end-to-end encrypted, because call setup passes through our server. Any call captioning would be offered as a separate feature chosen by the user.
Primary account and message data is currently located in the EU (Ireland), behind transport encryption and access controls. Separate services handle media, call relay, speech recognition and speech synthesis, sign-in email, notifications and website hosting; some of them operate in the United States. The store or payment service handles payment; Espero receives entitlement and transaction status, not card details. Media, logs and provider copies do not necessarily reside in the same country as the primary database. Transfer mechanisms are in sections 4 and 9 of the Privacy Policy.
Report a security issue to hello@espero.live before disclosing it publicly. We will assess the report and respond as appropriate.
Transport and provider-managed storage are encrypted, but ordinary cloud chats are not end-to-end encrypted because translation requires server-side processing. This description limits our claims to protections supported by the actual product and the providers' documentation.