The origin of the luminosity of superluminous supernovae (SLSNe) is an unresolved mystery, and a number of very different physical scenarios (including energy injection from magnetars, collision with a dense circumstellar medium and pair instability-induced explosions) have been invoked. The application of polarimetry to normal SNe has been shown to probe the three-dimensional structure of exploding stars, providing clues to the nature of the explosion mechanism. We report imaging linear polarimetry observations of the Type I SLSN 2017egm, in the galaxy NGC 3191, conducted with the Liverpool Telescope and the RINGO3 instrument. Observations were acquired at four epochs, spanning 4 - 19 days after light-curve maximum, however, polarization was not detected at a level of $>3sigma$. At +7 and +15 days, and in the average over all epochs, we find a possible polarization signal, detected at a significance of $approx 2sigma$ in the blue channel. This signal is seen, primarily, in the Stokes $q$ parameter, with a corresponding polarization angle consistent with the orientation of the spiral arm in proximity to the position of SN 2017egm. We interpret this as indicating that any polarization, if present, originates from dust in the host galaxy rather than being intrinsic to the SN itself. Despite its apparent peculiarities, compared to other Type I SLSNe, the polarization characteristics of SN 2017egm are consistent with the previously reported low polarization of other SLSNe of this variety.