First of all, the effects of these drugs have already been studied for many years. Modafinil was first experimentally used in the treatment of narcolepsy in 1986, so over 20 years ago.
Second, what do you mean proven beyond doubt? Testing these drugs is a statistical process, so what kind of confidence interval are you willing to accept? Keep in mind, that becoming more confident in your statistics means more testing, which makes the drug more expensive to bring to market and takes longer as well. Even then, a few drugs are going to pass the FDA approval process and cause unintended side effects, and people are going to argue for an even longer approval period.
You have to consider the opportunity cost of keeping these drugs in limited circulation. Most people don't take this into account, and there's plenty of research that shows that the FDA's requirements are actually probably too stringent, because government bureaucracies tend to be too risk averse. I think it takes an average of 12 years for a drug to pass FDA inspection and get to market, so imagine if it was 20 years. This means that during those extra 8 years, people weren't able to reap the benefits of a drug like Modafinil to stay awake. How many driving deaths could be prevented during those 8 years if people weren't driving while sleepy? Or what about the productivity gains from scientists or hackers that can get away with 4 hours of sleep a night instead of 8, without reducing their cognitive abilities?
I'm not saying there couldn't be unforeseen nasty side-effects of using the drugs fourty years from now, but given the evidence it seems like it's probably a negligible risk for most people.