DeepSeek — privacy & data policy

Provider: DeepSeek. PromptPrivacy score: 56/100. Last verified November 01, 2026.

What happens to your data?

DeepSeek collects your account information, chat prompts, uploaded files, and technical usage data, storing and processing them on servers in China. This data is used to provide chatbot outputs, improve services, and train AI models, and it may be shared with corporate affiliates and service providers. You can view or delete your chat logs in settings or email support to request data deletion or opt out of model training.

Does it train on your data?

Trains on your content by default, with an opt-out. DeepSeek uses user prompts, inputs, chat history, and diagnostic data across devices to develop and train machine learning models and algorithms, though users can submit a request to opt out.

Can you opt out?

Yes. DeepSeek lists the right to opt-out of using your personal data for training models or optimizing technologies, which users can exercise by emailing privacy@deepseek.com.

Data retention

DeepSeek retains personal data, including user inputs and account details, for as long as you keep an active account or as needed to satisfy legal and business obligations. When no longer required, data is deleted, erased, or anonymized.

Encryption & security

DeepSeek states that it maintains commercially reasonable technical, administrative, and physical security measures, but the policy does not specify encryption in transit (TLS), encryption at rest, or third-party certifications such as SOC 2 or ISO 27001.

Is it safe for work?

Not recommended for confidential or sensitive business data. User inputs are processed and stored on servers in China, content is used for model training by default, there is no disclosed SOC 2 or ISO certification, and the policy does not outline business-tier data protection agreements or enterprise administrative controls.

Source policy: https://cdn.deepseek.com/policies/en-US/deepseek-privacy-policy.html