The concern is of the association of the data that could be made. That's one part of it. As I mentioned for datasets, it's also this idea of collecting and using this kind of information in order to then try to engage in predicting criminality or predicting threat assessments. The ability to do that has not been proven. In fact, as we know, there are concerns around biases in those kinds of algorithms and that kind of analysis.
We've seen in the past, under national security investigations, that this idea of guilt by association is enough to engage in very harmful investigations.