I'm not talking about just any consensus, but scientific consensus, i.e. one arrived at through the scientific method. So far, it's only been broken when it had low confidence to begin with. Otherwise, it's only been refined. In both situations, this typically happens when there's new information.
For some intelligence to conclusively conclude that multiple scientific beliefs, held with a high degree of confidence, are wrong would be astonishing. It would mean either that the scientific method is bad, frequently applied incorrectly, or that conclusive relevant evidence is frequently ignored. Whichever of these is the case, it would also need cast a doubt on the intelligence's own conclusion.