Robust regression approaches enhance outlier detection by presenting more robust parameter estimates and by offering diagnostics and tools to spot influential observations through resistant estimation, residual analysis, M-estimation, etc. This way, you can find your outlier.

As Anthony Bagherian said, regression models can offer more tools than looking at the individual populations only (dependant and independent variable populations). A regression model can tell you if a sample does not have the expected covariance between variables, if it is not well explained by the model (high residual), or other metrics as Hotelling's T2, leverage... that might point out a sample as outlier.

Thanks Jokin Ezenarro and you are correct