The risk never goes down to zero. With driving, there are the abovementioned vehicle failures, there are also road conditions (obscured signs, broken traffic lights, etc) and obstacles (deer, moose, fallen objects, potholes etc); there are driver's decision errors (false assumptions of other's action and misjudging the gap to other vehicles or their speed being the most common), performance failures (overcompensation, panic/freezing, etc) and non-performance failures (fatigue blackouts, heart attacks, etc). Accidents, directly related to rules violations, have only a modest share: too fast for conditions 8.4%, too fast for curve 4.9%, illegal maneuver 3.8%, following too closely and aggressive driving 1.5% each (the percentages reffer to the number of total driver's error-related accidents, not to the total number of accidents). Same with oxygen. No matter how good you are with rules you can get distracted, get a wrong piece of equipment, drop the tank while moving, etc.
I never suggested the risks of either ever got to zero - but if you follow the rules/guidance, then it does get pretty close to zero. Even sitting in my chair in the office carries risk (I could choke on my coffee, have a heart attack or get a paper cut!). Nothing is every 100% risk free - it is however in all of our best interests , if we want to extend our lives, to try to tip the scales as much in our favour as possible.
I don't know your accident stats but I will bet they are based on millions of hours driven. So the chance of any individual having an accident is very near zero (when playing by the rules). I have to say that I would include driver decision errors, performance errors within driver errors as (staying well enough within the limits) they are ones that can often be mitigated by behaviour. There is an old driving adage over here "Treat every other driver on the road as an idiot until proven otherwise" - follow that and false assumptions of other drivers/misjudging of gaps etc have a tendency to move towards being non issues.
Even taking outside influences in to account, if you are driving within guidelines then road conditions/ obstacles/ performance failures do not lead to accidents - the driver is in the zone that they can take avoiding actions in time. We know that driving at reasonable speeds and adopting reasonable practises that most trips will be uneventful. Drive at over the speed limit, pushing the tyres to their grip limits in corners etc will move that "uneventful" into more than likely a few sketchy moments. Push even harder and it becomes increasingly likely that you will have an adverse outcome. Knowingly pushing your behaviour towards the limit or over the "line in the sand" increases your risk exponentially.
Accidents WILL happen and I acknowledge that (and have never denied it) but knowingly flouting safety guidelines for oxygen to me seems like a Darwin award application. Knowingly using equipment beyond that which it is expressly designed for with high % oxygen (such as Mac64's example) is taking big chances. For example he quoted the flashpoint of oils as 800f - that will be in 21% O2. Move the O2 percentage up to the high end of the scale (such as the 80% deco mix he quoted) and I am quite sure that the fuel (oil in this case) will have a far higher propensity to burn.