Does DeepSeek Train on API Data? A Policy Investigation

The Question

I use the DeepSeek API. Will my data be used to train their models?

The Answer

The Open Platform Terms are silent — they don’t say yes, and they don’t say no. Unlike OpenAI or Anthropic, which explicitly commit to not training on API data.

What Should I Do

1. Assume it does train — if the terms don’t say they won’t, don’t assume they don’t
2. Don’t send sensitive data via API — especially user privacy or trade secrets
3. Use local deployment instead — DeepSeek open-sources model weights (MIT License); run locally to eliminate the risk
4. Ask DeepSeek to clarify — privacy@deepseek.com / api-service@deepseek.com

Evidence

1: General Terms allow training, with opt-out (Web/App)

Source: DeepSeek Terms of Use — Section 4.3

“Under the premise of secure encryption technology processing, strict de-identification rendering, and irreversibility to identify specific individuals, we may, to a minimal extent, use Inputs and Outputs to provide, maintain, operate, develop or improve the Services or the underlying technologies supporting the Services.”

“If you refuse to allow us to process the data in the manner described above, you can opt out by turning off ‘Improve the model for everyone’.”

→ Web/App users: data used for training by default, but you can opt out.

2: Privacy Policy explicitly mentions training

Source: DeepSeek Privacy Policy

“To improve and develop the Services and to train and improve our technology, such as our machine learning models and algorithms.”

User right: “the right to opt-out of using your Personal Data for training our models or optimizing our technologies.”

→ Training use is documented in writing, and opt-out rights are granted.

3: API terms lack equivalent commitment (key finding)

Source: DeepSeek Open Platform Terms of Service

Full-text comparison with the general Terms:

→ The API terms neither confirm nor deny data use for training. No commitment = untrustworthy.

4: Industry comparison — competitors commit

→ DeepSeek’s silence is not industry standard — it lags behind competitors.