We’re considering the DeepSeek API for an internal prototype, and some prompts may contain snippets of non-public source code. Before we send anything, I want to understand the practical privacy questions to check rather than rely on a general statement that an API is suitable for business use.

For teams that have reviewed DeepSeek’s API data retention terms, what did you verify about logging, retention periods, and whether submitted data can be used for model improvement? I’m also interested in how you handle sensitive inputs when the policy or account settings leave room for interpretation.