This paper investigates the leaderless consensus control for fractional-order multi-agent systems (FOMASs), in which the sensors are subject to unknown measurement sensitivity and the actuators are subject to deception attacks, respectively. Due to the special nature of fractional calculus operations, the control design for FOMASs with the unknown sensitivity and attacks is more challenging. In order to solve these difficulties, some auxiliary variables are constructed deliberately with which the distributed adaptive control scheme is proposed. Meanwhile, we design a fractional-order filter to avoid the problem of “complexity explosion”. Compared with the existing filters, the proposed filter can confirm that the filter errors are asymptotically convergent and the unknown upper bound of fractional-order derivatives of virtual control signals is estimated effectively. From the rigorous theoretical analysis, it is proved that under the proposed adaptive control, the asymptotic consensus can be achieved in presence of the sensors with unknown measurement sensitivity and actuator attacks. Finally, two examples are applied to verify the feasibility of the presented control scheme.