Gas selection is irrelevant, 100 000dives/1 fatality in rec diving and 6000-10000 dives/1 fatality in CCR. These are the odds.
This is an oranges and apples comparison. If you normalize for risk exposure of dives (e.g., PrT — avg depth x square root of time) I assume the split would be different. Tech dives are inherently more dangerous than NDL dives in OW, this is why tech diving requires equipment, procedures, and training to mitigate the risk.
Deep diving on air on a single tank is more risky than deep diving (to the same depth for the same time) with appropriate equipment, gases, and procedures.
Risk is not zero for either option but it’s lower in the second one. This also doesn’t mean that diving with the first option will result is an accident every time, as testified by team deep air in this thread. But let’s please stop asserting that the risk is the same just because it worked in the past.