N.B. A potentiometer's power rating assumes a uniform current through the whole track. If you are only using 10% of the track length it only has 10% of the power rating. Divide the wattage by the total resistance and take the square root to get that max rated current. If you exceed that current through any terminal (including the wiper) you'll burn out the pot sooner rather than later. Assuming a 1W rated potentiometer, with a 100K pot, its good for max. 3.16mA.
With 47V across it your pot only has 0.47mA goint through its track. That isn't much to supply base current for your triple Darlington emitter follower, especially as one generally doesn't want to draw more than 10% of the track current at the wiper. If you switch to a 10K pot, its going to have 4.7mA through it and is good for an abs. max current of 10mA. The down side is the greater load on the Zener shunt regulator circuit, as the 2K2 resistor only provides 8.6mA. Looking at the OnSemi 1N4756A datasheet, its tested at a nominal 5.5mA Iz, and 3.9mA remaining for the Zener will be good enough.
When you rapidly turn the output voltage down, the previous voltage the output capacitor was charged to causes Vout to exceed Vwiper. If that's by more than 15V or so, all the B-E junctions will Zener, and pop goes your pot and probably the weakest transistor as well. The easiest fix is to put a diode in series with the pot wiper to block reverse current. As Andy points out above, you'll also need a B-E resistor for each transistor to handle its worst case C-B leakage current, and with the wiper diode in place, the B-E resistor also prevents the B-E junction zenering.
You've still got the problem of protecting the transistors against the surge current during startup, or if the load draws too much current. However adding effective current limiting to the design becomes quite tricky, as it would need to be fold-back limiting to protect the transistors effectively, and it rapidly turns into a full-blown high current high voltage bench PSU design exercise which is far from being beginner friendly.
As an alternative, if all you need is variable 'raw' DC for amplifier testing, consider scrapping everything to the right of the fuse and simply putting a variac on the AC input!