Although the distributed bad data detection algorithm can eliminate the influence of bad values for the least square estimation performence, there are still security vulnerabilities in the existing distributed bad data detection algorithms. Motivated by the above-mentioned fact, this paper proposes a hidden false data injection (FDI) attack method to degrade the performance of the distributed least square estimator, where the hidden attack method is designed based on system information of some nodes. This method not only makes the FDI attack signal unable to be detected by the existing distributed bad value detection methods, but also changes the estimation result with a preset offset. Finally, the IEEE 118-Bus power system is used to illustrate the hiddenness and effectiveness of the proposed methods.