Facebook's role is to escalate, not to intervene, and even that escalation process is done by a human. It seems very unlikely that it would ever result in a sit

Facebook's role is to escalate, not to intervene, and even that escalation process is done by a human. It seems very unlikely that it would ever result in a situation where someone gets involuntarily hospitalized because some machine learning algorithm had a bug. Per TFA they don't actually track the outcomes of their calls because that would be too much of a privacy violation. I'd expect the resulting lack of feedback to be problematic. Also, I'd say that escalating to the people authorized to use force is a form of intervening.