In this talk, I will introduce an inverse source problem for the multi-term time-fractional stochastic differential equation driven by a fractional Brownian motion with Hurst parameter $H\in (0,1)$. For the direct problem, the multi-term time-fractional stochastic differential equation is proved to admit a unique mild solution. For the inverse problem, we prove that the variance of the boundary data over the whole time domain can be used to uniquely determine the modulus of the Fourier modes of the diffusion coefficient involved in the random source. To further get the diffusion coefficient from its phaseless Fourier modes, the PhaseLift method combined with spectral cut-off technique is used to recover the modulus of diffusion coefficient numerically. Several numerical experiments are also reported to demonstrate the effectiveness of the proposed method.