This work focuses on a vulnerability of hash functions due to sloppy usages or implementations in the real world. If our cryptographic research community succeeded in the development of a perfectly secure random function as the random oracle, it might be broken in some sense by invalid uses. In this paper, we propose a new variant of the random oracle model in order to analyze the security of cryptographic protocols under the situation of an invalid use of hash functions. Our model allows adversaries to obtain contents of the hash list of input and output pairs arbitrarily. Also, we analyze the security of several prevailing protocols (FDH, OAEP, Cramer-Shoup cryptosystem, Kurosawa-Desmedt cryptosystem, NAXOS) in our model. As the result of analyses, we clarify that FDH and Cramer-Shoup cryptosystem are still secure but others are insecure in our model. This result shows the separation between our model and the standard model.