DeepL — privacy & data policy
Provider: DeepL. PromptPrivacy score: 42/100. Last verified November 01, 2026.
What happens to your data?
When you submit text or documents to the free DeepL service, they are encrypted in transit and processed on DeepL's servers in Europe to produce your translation. DeepL retains your input, translations, and corrections for a limited period to train and improve its neural networks. Personal data is expressly prohibited in the free tier, and immediate deletion upon translation is available only to paid DeepL Pro subscribers.
Does it train on your data?
Trains on your content by default. When using the free version of DeepL Translator or DeepL Write, all submitted texts, uploaded documents, translations, and user corrections are processed and stored for a limited period to train and improve DeepL's neural networks and algorithms. Only paid DeepL Pro subscriptions guarantee that submitted content is never used for model training.
Can you opt out?
Not specified. The policy states that content submitted to the free service is processed to train neural networks, but it does not specify any self-serve opt-out setting for free users. To prevent user content from being used to train models, users must upgrade to a DeepL Pro subscription.
Data retention
For the free version of DeepL, submitted texts, uploaded documents, and translations are retained for an unspecified 'limited period of time' to train models and improve algorithms. Temporary document translation metadata and full IP addresses are retained for up to 14 days before automatic deletion. Account profile information is stored for the duration of the account.
Encryption & security
Data is encrypted in transit between the user's end device and DeepL's servers, ensuring intermediaries like Cloudflare only have access to metadata. The policy does not specify whether data is encrypted at rest, nor does it mention independent security certifications such as SOC 2 or ISO 27001.
Is it safe for work?
No, the free DeepL service is not safe for business use. DeepL's terms explicitly prohibit using the free translator for texts containing personal data of any kind, and all submitted content is retained and used to train DeepL's AI models. Organizations handling confidential or personal data must use DeepL Pro, which includes a Data Processing Agreement (DPA), prompt text deletion after translation, and an exemption from model training.
Source policy: https://www.deepl.com/en/privacy