YES - Tabnine's private code model is completely safe. The private code model that is created by the learning algorithm is strictly exclusive: it is made and delivered only for approved team members. Nobody else has any access to it or the ability to run it, as it is protected by an identity verification process that ensures the user indeed belongs to this particular team. In addition, all communication and delivery of the learnings and model to and from Tabnine's secure cloud is encrypted and is done using secured channels. Finally, the learnings data used for creating and updating the model are immediately deleted after the team learning algorithm updates the model.
Enterprise customers also have an option of self-hosting, where your private code model is created and updated in a self-hosted Tabnine server. This means that no contextual information or any code fragment ever leaves your organization. If this is relevant for you, please click here.