I'm somewhat concerned about the ethical implications of this beyond the scope of just car crash reporting. While impressive, the nlp model will have to be tune

I'm somewhat concerned about the ethical implications of this beyond the scope of just car crash reporting. While impressive, the nlp model will have to be tuned by someone, someone with biases the model will inevitably inherit. Seems to have obvious potential as a propaganda tool, which might be used by the 'good guys' today, and be used by malicious actors tomorrow.