Recent studies reported a differential multisensory integration (MSI) in autism spectrum disorder (ASD). Much of the research on MSI differences has focused on how visual stimuli influence speech processing. The present study takes a reverse perspective. We investigated if speech processing can affect the construction of low-level visual representations in autistic individuals when they detect a visual stimulus under the continuous flash suppression (CFS) effect. The results showed that nonautistic participants benefit from speech processing during visual object detection when the linguistic stimulus is congruent with the masked visual object. This effect was not observed in autistic participants. In turn, the reaction time (RT) of correct answers was significantly lower in the nonautistic group than in the ASD participants. This indicates that MSI between linguistic and visual stimuli is affected in both directions. Our findings are interpreted under the embodiment cognition framework and its relationship with ASD.