>>41048382Theoretically, yes. It very well may have been a major factor in making the filter go absolutely berserk at one time. It's easy to point fingers to the association death problem being the primary cause of that, which is likely the fault of the filter AI's unmanaged growth for several months. However, in response to it making the service essentially unusable for a small period, my assumption is that a leash may have been put on it now. It still hits false positives to be sure, but it seems much easier to beat your head against it until you receive satisfactory results at this point. Bypassing methods which I will not discuss do tend to work if you try hard enough. That said, if the filter does get upgraded like
>>41049308 suggests, I honestly can't imagine it not doing the same thing it did before. I do not believe that a filtering AI would be able to properly function without being either going into absolute tyranny over anything even remotely against TOS or being a wet blanket like it kind of is now.
