That could be done before, but now you could automatically flag and harass people based on the transcriptions and a lot of people would think it is a quite robu

That could be done before, but now you could automatically flag and harass people based on the transcriptions and a lot of people would think it is a quite robust system, or will in the future when the only one they actually confide in is a chatbot.