Attribute Based Encryption (ABE) has found enormous scope in data confidentiality and fine-grained access control of shared data stored in public cloud. Classical ABE schemes require attaching the access policy along with the ciphertext, where the access policy describes required attribute values of a receiver. As attributes of a receiver (i.e., user) could relate to the identity of users, it could lead to reveal some sensitive information of the ciphertext (e.g. nature of plaintext, action sought from of receiver) for applications like healthcare, financial contract, bureaucracy, etc. Therefore, anonymizing attributes while sending ciphertext in use of ABE schemes, known as Anonymous ABE (AABE), is a promising primitive for enforcing fine-grained access control as well as preserving privacy of the receiver. In ASIACCS 2013, Zhang et al. proposed an AABE scheme using the match-then-decrypt [1] technique, where before performing decryption, the user performs a match operation that ensures a user whether he is the intended recipient for the ciphertext or not. We found that Zhang et al.’s scheme [1] is not secure, in particular, it fails to achieve receiver’s anonymity. In this paper, we discuss the security weaknesses of Zhang et al.’s scheme. We show that an adversary can successfully check whether an attribute is required to decrypt a ciphertext, in turn, reveal the receiver’s identity. We also suggest an improved scheme to overcome the security weakness of Zhang et al.’s scheme.