We design an efficient cryptographic method for verifying ML fairness with respect to data received online during deployment.