Why would this apply only to late game tech research? Isn't any research, by definition, cutting edge when it is occuring? I can see the later game research MAY be more potentially devastating, but doesn't research at any time have the chance to go wrong?
Are there different things you might put in place that could reduce or minimize that damage? The FDA inhibits the speed at which medicine gets to market, but it also decreases the likelyhood of some really bad meds affecting the greater population as a whole.