Tabnine — privacy & data policy

Provider: Tabnine. PromptPrivacy score: 85/100. Last verified October 18, 2026.

What happens to your data?

Tabnine uses code snippets from your local environment strictly as short-lived context to generate completions and chat answers. Once the output is generated, the code is immediately discarded and never stored or shared with third parties. Only system operational metrics and one-way hashed non-identifiable telemetry are logged for up to a week.

Does it train on your data?

Tabnine enforces a strict 'no-train-no-retain' policy and does not train its AI models on user code. Its code completion and protected chat models are trained exclusively on permissively licensed open source code. Private fine-tuned models are trained only on private code specifically authorized by an enterprise team and remain isolated to that customer.

Can you opt out?

Not specified. Tabnine operates with zero retention and no model training on user code by default, eliminating the need to opt out of training, and the text does not describe opt-out options for operational telemetry.

Data retention

User code sent as context for completions or chat is processed ephemerally and deleted immediately once an answer is returned. Tabnine does not store user code. System operational metrics and logs from self-hosted clusters are retained on Tabnine's servers for one week.

Encryption & security

Not specified.

Is it safe for work?

Yes. Tabnine is designed for professional and enterprise development, featuring a strict zero-retention, zero-training stance on user code, no third-party API exposure, and deployment options for on-premises, VPC, and air-gapped networks.

Source policy: https://docs.tabnine.com/main/welcome/readme/privacy.md