Covert communication hides signals with a low probability of being detected by a warden, achieving a high level of security. This paper focuses on the multiuser covert system where K pairs of cover users communicate in the same frequency and the warden monitors all transceivers’ transmission. Without relying on external artificial noise or public links, we wonder whether covert performance can be improved by covering each other. Considering the probabilistic co-existence of users’ transmission, a detection framework based on a multi-hypothesis test is formulated to describe the \(2^K\) combinations of users that transmit simultaneously. For analytical simplicity, we decouple the joint K-user test to K singer-user tests, converting the multi-hypothesis test into multiple binary hypothesis tests. We derive the relative entropy between two complex Gaussian mixed models as the low bound of the detection error probability. We analytically and numerically show that the relative entropy is not a monotonic increasing function of the transmit power, but grows in fluctuation. The numerically results further demonstrate the multi-user system achieves better covert throughput than the single-user system, revealing the uncertainty from the detection of multiple users can enable covert transmission.