Fault attacks are a common threat for embedded secure implementations. Among the various kinds of countermeasures proposed so far, the principle of infective computation seems to be one of the most efficient ways to counteract this threat. However, each and every original infective countermeasure suggested for asymmetric cryptosystems has been broken. Nowadays only two propositions for symmetric ciphers are still believed to be secure. Our paper presents the first attacks on both infective symmetric implementations, thus proving that these propositions rely on incomplete security analyses. By breaking the two last surviving infective methods, this paper shows once again that it is very difficult to design a secure infective countermeasure.