@benjojo call me cynical (I am), a lot of this stuff really seems more like a form of "back channel peacocking" than reality. When they tell you their bot tried to put a hit on someone, that's essentially bullshit, but framed to "ooh... it's all so deep". This is no different.
They don't need a policy manager for this, questions around these aren't happening enough to need a policy manager, but how fancy pants do they look if they say they are?!