Although web companies claim to support an inclusive content moderation policy and to be committed to combating online hate, they are regularly accused of censoring minority communities. This article examines how moderation features of digital social media (particularly Twitter) affect the visibility regimes of LGBT and of sex workers online. The study uses digital methods for tracking and collecting deleted tweets, to formulate hypotheses about what prompted the deletions leading to excessive censorship. In particular, it explores the hypothesis that certain keywords referring to sexual and gender culture trigger automated moderation due to their offensive nature.