Comment by aarushyadav

When we launch, no, it will be cloud inference, but will be processed with encryption in hardware isolation and only retained and used for anonymized training with your consent. At some point in the future we may make models available for local inference though.