<p>Drawing on an intergroup contact framework, we investigated the effectiveness of a new human-robot interaction (HRI) procedure—<i>human-robot (fast) friendship (HRF)</i>. HRF is a self-disclosure-based interaction between a human participant and a robot, designed to improve attitudes towards social robots and non-contacted secondary human outgroups (e.g., vegetarians, refugees, and individuals with disabilities). Study 1 tested HRF with a humanoid robot among university students (<i>N</i> = 103), and demonstrated HRF to directly and indirectly improve attitudes towards social robots through greater trust and positive emotions (sympathy and respect). HRF was also indirectly associated with more positive attitudes towards some secondary outgroups. Study 2 replicated the same procedure in a community sample (<i>N</i> = 148); participants in the HRF condition showed more positive attitudes towards social robots through increased trust and in turn sympathy, compared to both baseline (no contact) and mere exposure (being in the same room with the robot) conditions. In Study 2, we also observed indirect effects on attitudes towards some human outgroups through increased robot sympathy. Findings highlight the importance of friendship potential in HRI effectiveness and point to the potentially broader consequences of HRF in reducing generalized prejudice in humans.</p>