This is also one of the reasons why AI makes totalitarianism a viable strategy. In places like stalinist Russia, or maoist China, there was always a problem with finding enough ultra loyal, intelligent, uncorruptible servants of the state's power. There were genuine loyalists that the supreme leaders could count on but they never knew for sure which of their followers believed the big lie and which of them were just playing along for personal advantage or maybe even subscribe to a hidden agenda. Now once you have high confidence in an AI surveillance model or an AI enforcement model, you can just clone that ad infinitum.