It's tricky: there is often monitoring of interim results for exactly this reason, but there are some good theoretical reasons to think that halting a trial on the basis of good interim outcomes in the experimental group could bias the results.

As for the latter statement, there are no "guarantees" before clinical trials have been done (even after it can be iffy: see http://www.plosmedicine.org/article/info:doi/10.1371/journal....

I can't imagine how this argument would look in court.