Humanitarianism has long been thought of as the promotion of human welfare. In the context of peace and war, humanitarianism refers to the delivery of safety and protection to vulnerable populations. But humanitarianism has another history – one that cannot be understood apart from a history of experimentation, including experimental colonial and postcolonial endeavours in foreign territories and on foreign bodies to test new technologies and to make them safe for use by more valued citizens often located in metropolitan states. In this article, I explore how the United Nations High Commission for Refugees' (UNHRC) repatriation of Afghan refugees from Pakistan between 2002 and 2007 constitutes a case of humanitarian experimentation because of how the UNHCR deployed iris recognition technology to control the flow of refugees across the Afghan/Pakistan border. I argue that the UNHCR's use of iris recognition technology not only served to detect ‘deserving’ refugees, correct refugee movements, and innovate a manageable refugee. It also cultivated technological failures, corrected the technology's performance and linked local refugee databases to international biometric databases, all of which potentially introduced new insecurities into this border region and beyond. The effect of humanitarian experimentation in this case was to render the safety of refugees subordinate to the production of apparently safe technologies, so much so that refugees subjected to humanitarian experimentation in the world's dangerous ‘wild zones’ delivered ‘safety’ (by delivering presumably safer technologies) to citizens in the world's relatively secure ‘tame zones’ – not the other way around. In this way, the UNHCR's use of iris recognition technology made design safe for some citizens, but not for others.