Social media platforms constantly face decisions with no straightforward answer: should nudity policies bend for an image of historical significance? Should hate speech and violence rules carve out space for civilians documenting atrocities in conflict zones, content that may later prove vital for war crimes trials? These are instances of “borderline content” — material that sits in a zone of contextually contingent ambiguity, often described as “lawful but awful”. They are not a clear policy violation, but still raise serious questions of harm and social impact that platforms must adjudicate at scale.
This research examines what makes such content difficult to govern, through the interplay of three factors: the content itself (visual, textual, or audio features that approach a policy boundary without clearly crossing it — even supposedly well-defined categories like CSAM involve interpretive edge cases); the conduct surrounding it (patterns of behavior, coordination, or intent that turn otherwise permissible content problematic); and the context in which it appears (newsworthiness, artistic value, or cultural and legal setting, which can shift how the same content reads across jurisdictions).
Drawing on policy analysis, platform case studies, and interviews with content moderators and trust and safety practitioners, the research traces how borderline content moves through escalation processes and exposes a critical power imbalance: decision-making authority sits with Western, English-speaking executives, while reviewers with the local cultural and linguistic expertise to interpret these cases — often based in Majority World countries — have little influence over policy itself. The findings point toward more contextually responsive governance, better stakeholder coordination, and, centrally, restructuring decision-making to give Majority World expertise real authority rather than a consultative role — a challenge only growing more urgent as AI-generated content joins user-generated content in testing the boundaries of platform policy.
