Classic research on human factors has found that automation never fully eliminates the human operator from the loop. Instead, it shifts the operator’s responsibilities to the machine and changes the operator’s control demands, sometimes with adverse consequences, called the “ironies of automation.” In this paper, we revisit the problem of automation in the era of social media, focusing on privacy concerns. Present-day social media automatically disclose information such as users’ whereabouts, likings, and undertakings. Our review of empirical studies exposes three recurring privacy-related issues in automated disclosure: 1) insensitivity to situational demands, 2) inadequate control of nuance and veracity, and 3) inability to control disclosure with service providers and third parties. We claim that the “all-or-nothing” type of automation has proven problematic and that social network services should design their user controls with all stages of the disclosure process in mind.