Well going back to basics.
The problem here is the lack of information as to construction necessary to produce a predictable result.
So
From my perspective assuming that there is no light elements onside the room, and you are only using a single point source, then there is no expatiation of a light leak to begin with then the problems originate from that source.
Based on just this image, as to an expected result, there is a lack of required elements that is required beyond making a basic room, tossing in a light element, and tada you have a real world result
To guess once again I would assume as to expectations, with a single point source, was under lite so you ramped up the intensity and increased the the lighting volume to fill the volume using force beta.
Note : force beta is where a value is increased beyond it’s acceptable values (turning things past 11) creating an unexpected result so to figure out what is going on you should be providing a scene example and not just a static image.
That said we need to go back to rendering basics as to what is needed in the construct. Think like your making a cake where if you eliminate just one ingredient what you land up is anything but a cake.
Ingredients for our cake.
Camera. The camera is of course is what you rendering to and you can set it up to make use of whats called an in-camera effect that can produced the same results as you could with a post process volume. Nothing much to do here if using a stock camera but if you copy for some reason a camera from another project the settings of that camera could compound the problems
Light source. A light is a light and since UE4 uses real world lighting models then it should behave in a predictable manner as to it’s real world counterpart. It should be noted that by default a light element uses inverse fall off for shadows (Google it), which in most 3d apps is turned off by default so anyone new to lighting theory would assume that more intensity is required which in turn only makes the result worse.
Materials. This is where lighting behavior does it thing based on the nature of PBR based materials and since it’s based on real world light models is not the reinvention of the wheel so the expectations should be no different if you are rendering in any other 3d application except for value scaling and of course give a material function a different name (roughness for example is glossiness in 3ds Max)
Looking at you image sample using the wrong material type could be your problem, which is usually the same problem in “any” 3d application, so I would suggest trying the free automotive material as it has a very high reflective quality which in turn “will” ramp up the indirect lighting levels based on the total number of bounces.
Back to lighting theory what your looking for is a “balance” lighting result so you have a solid base from which to work from and the image you have proved does not even come close as to an expected result from which to work from.
Example.HDR GI - YouTube
Guess how many lighting elements there are and what kinds ?
So looking at you image as making a cake maybe your missing some eggs?
Hope that helps and at some point answers you question