Covariance shaping least squares estimator is a new kind of linear estimator based on covariance shaping in quantum signal processing (QSP) framework, it is a biased estimator directed at improving the performance of traditional least- squares (LS) estimator. In this paper, we discuss the covariance shaping least squares (CSLS) estimator, and apply it in multi-user detection that leads to a new detection algorithm, and then we employ it in the detection of MIMO systems. The numerical results in synchronous DS-CDMA communications system show that the performance of covariance shaping multi-user detection which depends only on the users' signatures is better than that of traditional MF and decorrelation detection, and is an approach to the MMSE detection method. The results in MIMO systems also show that the performance of system is better than that of detection algorithm with Zero-forcing and MMSE solutions.