What is DeepSeek's policy on data deletion upon user request?

DeepSeek’s data deletion policy allows users to request the removal of their personal data in accordance with privacy regulations like GDPR and CCPA. When a user submits a deletion request through an API endpoint or via email, the system initiates a verified process to erase identifiable information such as account details, usage logs, and user-generated content. For example, a developer using DeepSeek’s API could trigger a deletion workflow by sending a DELETE request to /users/{id}/data with proper authentication. The process typically completes within 30 days, though some backups may take up to 90 days to purge due to technical constraints. This approach ensures compliance while maintaining system integrity.

Technically, DeepSeek handles data deletion by tagging records for irreversible removal rather than immediate physical deletion. Data is stored in distributed systems like AWS S3 for files and MongoDB for structured data, with deletion jobs running nightly. For encrypted data, decryption keys are destroyed to prevent recovery. Developers should note that anonymized analytics data (e.g., aggregated API usage patterns) is exempt, as it doesn’t identify individuals. The system also provides audit logs through an admin API (e.g., GET /audit/logs?type=data_deletion) to track deletion events. This layered approach balances compliance with performance needs, avoiding full database scans that could impact live services.

Exceptions exist for legal holds or ongoing fraud investigations. For instance, if a user account is involved in active litigation, DeepSeek may retain specific records while deleting non-essential data. Developers integrating with the platform receive webhook notifications (e.g., data_deletion.partial) in such cases. Users are informed via email about what data was kept and why, with timestamps referencing legal documentation. This granular control lets technical teams adjust retention policies programmatically while maintaining transparency. For example, a PATCH request to /data_retention/rules could exempt certain datasets from deletion based on predefined legal criteria.