Tabnine — privacy & data policy

Provider: Tabnine. PromptPrivacy score: 85/100. Last verified August 26, 2026.

What happens to your data?

Your code context is processed ephemerally to provide code completions and chat responses, then deleted immediately. Tabnine never stores your code, never uses it to train AI models, and never shares it with third parties.

Does it train on your data?

Tabnine does not train its AI models on user code. Its code completion and protected chat models are trained exclusively on open-source code with permissive licenses. Any private fine-tuned models remain isolated and accessible only to your team.

Can you opt out?

Not specified. Opt-out from model training is not required because Tabnine enforces a strict zero-retention and no-training policy by default across all tiers.

Data retention

Code context sent for inference is processed ephemerally and discarded immediately after the server returns the answer. Operational metrics and system logs (which exclude code and PII) are retained for one week.

Encryption & security

Not specified. The policy notes that telemetry identifiers like email, hostname, and IP are one-way hashed, but it does not detail encryption in transit, encryption at rest, or security compliance certifications.

Is it safe for work?

Yes. Tabnine operates on a strict no-train-no-retain policy, never shares code with third parties, and provides enterprise options (on-premises, VPC, air-gapped) where no code or PII leaves the organization's environment.

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