Neither of the stories you site indicate the dogs gave false positives. One said the dogs can give false positives. The truth is, the dogs are 90% accurate. 9 out of 100 give a false positive.
Like you though, I am holding out hope we have one of the nine.
http://dogsdontlie.com/main/2008/12/cadaver-dogs-how-reliable-are-they-at-detecting-death/
"In other words, out of 100, there are 0-9 false positives and 0-25 false negatives. A negative predictive value describes the chance that, if a sample is not contaminated, the dog will correctly identify the sample as clear of human remains. The study quotes a negative predictive value of 90-100. This means only 0-10 clean squares are wrongly identified as contaminated by the dogs.
Accuracy is the degree to which the evidence presented by the dogs matches known information about which squares were marked. The accuracy of dog detection is presented as 92-100. This means that dogs correctly identify carpet squares as marked or unmarked in at least 92 cases out of 100. This is an impressive accuracy score.
In addition, I think it is important to consider that this is an experiment, not real life. In reality cadaver dogs are given more time to assess possible traces of human remains. Hence in a true police setting, cadaver dogs are more likely to give accurate information."