DeepSeek
DeepSeek Chat
Verified: 2026-06-22 Changed: 2026-06-23
F47 Pts
Privacy Grade Breakdown
Scores are calculated based on default weights. Visit the Methodology page for the formulas.
Default Training
F0/100
Opt-Out Ease
A+100/100
Data Retention
F0/100
Deletion Rights
A+100/100
Third-Party Sharing
B85/100
Human Review
F30/100

Policy Details & Nuances

Default Model Training
Yes

Free consumer chats on DeepSeek Chat are used to train and optimize models by default. API platform usage is excluded from training.

Opt-Out Mechanism
Yes

In consumer settings, turn off the 'Improve the model for everyone' toggle to opt out of the model training program.

Data Retention & Deletion

Retention Period

Retained as long as necessary for business purposes. De-identified or aggregated data may be retained indefinitely.

Deletion on Request

Yes

You can delete chats from your history or request full account deletion. Deletion requests are processed per local regulations (such as CAC in China and GDPR in Europe).

Third-Party Data Sharing

Shared with affiliates, infrastructure providers, and authorities under Chinese cybersecurity laws and international compliance rules.

Human Review of Chats
Yes

Conversations may be reviewed by human annotators for compliance with Chinese safety laws (CAC) and model optimization.

Context & Variations

Regional Variations

Domestic Chinese users are subject to CAC regulation. International users have separate privacy policies and data hosted outside mainland China (e.g., Singapore).

Children's Data Policy

Limited to users 18 years or older. Underage accounts are terminated upon discovery.

Enterprise vs. Consumer
Yes

DeepSeek API platform usage does not train on query data by default. It offers temporary data caching (up to 30 days) for abuse detection and complies with standard enterprise security expectations.