Once each chip reaches a critical threshold for temperature the leakage and thus heat becomes a run away train, and its exactly why Intel has such tight voltage control on their newer smaller process chips, its what causes tunneling and electrostatic discharge, the inrush of current as the gates open and close and capacitive build up will destroy a chip, and exactly why temperatures, caused by either poor cooling, from the AIO, die to IHS interface ( I know its soldered), the IHS itself cause this runaway stability effect.
At the end of the day he **SHOULD** have enough cooling capacity to keep it reasonably cool assuming his room isn't 30C or hotter and he has decent airflow to the radiator.
But until its verified that the VRM's are the issue and not the cooling in one function or another, I would assume for safety that its damaging the die when it fails as it may have reached the threshold for heat and leakage before it becomes a cascade effect, and the reasonable course is to, as has been stated to lower the speed and voltage, and if needed I would get a laser thermometer to check the VRM's, and also the area around the socket.
When you pulled the block off was there any noticeably thicker areas of TIM? Most of the time you should have the ring of where the edge of the IHS was, and barely enough left in the middle to show a haze, and you should be able to read the lettering on the IHS. If not you may benefit from lapping the IHS.
