Recent work in social epistemology has discussed obligations to engage with challenges to our beliefs like climate change denial or anti-vaccine sentiment, and the potential benefits and dangers for both the engager and the engaged from doing so. The spotlight being trained here, however, has elided a key issue: the possible risks from engaging for third-party observers, not merely the engager and the engaged. In this paper, I argue that not only are these risks an underappreciated aspect of engaging that should be discussed, but also their neglect is especially concerning as the potential negative epistemic fallout threatens to overwhelm any possible benefits that may be gained from engaging, regardless of how the engagement actually goes. I close by drawing out the theoretical and practical implications from this and sketch a few strategies to conceivably avoid said risks.