Attribute-Based Encryption (ABE), a special type of public key encryption, efficiently shares sensitive data with fine-grained access control. ABE can be classified into two types: Ciphertext-Policy ABE (CP-ABE) and Key-Policy ABE (KP-ABE). However, the securities of most presented ABE systems were reduced to the q-type DBDH (Dicisional Diffie-Hellman Assumption) assumptions, which are stronger than the DBDH assumption. So, the abovementioned ABE systems become insecure if DBDH is proved to be insecure. We propose a new ABE framework, called security-level switchable ABE (SLS-ABE). In SLS-ABE framework, a series of ABE systems can be generated and their securities are reduced to a k-BDH assumption family proposed by Benson et al. The k-BDH assumption family has the following properties: 1) any assumption in the k-BDH assumption family is associated with a parameter k, and the assumption becomes strictly weaker as the parameter k increases. 2) the 1-BDH assumption is proved to be equivalent to the DBDH assumption. So, all the k-BDH assumptions where k > 1 are weaker than DBDH assumption. We apply the technique of Benson et al. to construct ABE on k-BDH assumption, furthermore, we design a new framework to support the flexible switchable security-level for users. Concretely, the master public key, the master secret key and core keys issued by the system are constant. A User can generate different security-level public key/secret key pairs if it holds the core key. We propose a public key forgery attack model (PKFA) to capture the behaviors of adversary for generating a forged public key. We formally prove the selective-CPA security and PKFA security of our ABE systems. We compare the performances of our systems with Waters’ ABE systems.