Recent research has problematised online harassment as a risk in social media research. Scholars have since tried to address this through protective mechanisms in research processes, above all in antagonistic spaces such as far-right online groupings. This paper addresses a comparatively little explored area: risk identification and research ethics in ambiguous digital spaces. It draws on an ethnographic case study of a climate change sceptical Facebook group that showed antagonistic tendencies in later stages of observation. It discusses how such groups may pose risk to researchers where they are part of the demographics typically targeted in them: scientific elites, female, millennial, non-denialists, and international. The paper critically discusses researcher-sensitive approaches in research ethics, including contextual integrity and non-disclosure of research intent, in terms of who these processes protect or expose given exclusionary right-wing gatekeepers and their creative mediated processes aimed at harassing outsiders. It will argue in favour of benefits gained from a) an ethical screening stage that considers latent aspects of ideological positioning such as moderation practices and visual vernaculars, and b) multimodal ethical assessments employing feminist and minority ethics of care as part of the vulnerability triad.