AI translation and GDPR: how to recognise a secure service

Official documents contain personal data. So what matters is not only how well an AI tool translates, but also where and how it processes the data. The GDPR sets clear standards for this.

A translator’s tool: scan → precise translation in the original layout. Clients, signature and fee stay yours Try free — 5 pages

Six features of a GDPR-compliant AI service

  • Contract: the provider offers a data processing agreement under Art. 28 GDPR.
  • Location: data is processed within the European Economic Area; there is no transfer to third countries (Art. 44 et seq. GDPR).
  • Purpose: content is not used to train models.
  • Retention: documents are stored only as long as the translation requires (Art. 5(1)(e) GDPR).
  • Security: data is encrypted (Art. 32 GDPR).
  • Transparency: all sub-processors are disclosed (Art. 28(2) and (4) GDPR).

Many free consumer translation services do not meet these features, or cover them only in their terms of use. A tool for translating official documents should meet them from the start.

How Lingverra meets these features

  • Contract: the data processing agreement under Art. 28 GDPR is published and applies automatically to every account — no separate request needed.
  • Location: servers are hosted by netcup in Germany and Austria. AI translation runs via Opper (Sweden) on Amazon Bedrock in Sweden; a contractually bound fallback channel is in Frankfurt (AWS). Processing outside the EEA is technically blocked.
  • Purpose: your documents are not used to train models; Opper keeps no copies of requests; on Amazon Bedrock, requests and responses of the Claude models used are not stored and are not passed to the model developer.
  • Retention: source and translation are deleted automatically 72 hours after delivery — or manually at any time.
  • Security: every document is encrypted with AES-256-GCM using a separate key per document, destroyed when the document is deleted; the keys are protected by AWS KMS in Frankfurt.
  • Transparency: the full list of sub-processors with purpose, location and contractual basis is public.

For your records

You can link to or file the data processing agreement and the list of sub-processors directly — for example in your record of processing activities, or when a client asks how their documents are handled. Both are listed below under sources.

Read next: AI translation for sworn translators — opportunities and limits

Read also: Where your client's document goes when you translate with Lingverra

Read also: Sworn translators and the GDPR: are you a controller or a processor?

Sources

Try it on your own document — 5 pages free. Lingverra takes the routine off your hands for pennies, while clients and fees stay yours.

Your first translation right away, no profile to fill in · data processed in the EU

Try for free

Discover Lingverra →

New articles first on LinkedIn →