Programming the machine: Gender, race, sexuality, AI, and the construction of credibility and deceit at the border

There is increasing recognition of the significance of the political, social, economic, and strategic effects of artificial intelligence (AI). This raises important ethical questions regarding the programming, use, and regulation of AI. This paper argues that both the programming and application of AI are inherently (cis)gendered, sexualised and racialised. AI is, after all, programmed by humans and the issue of who trains AI, teaches it to learn, and the ethics of doing so are therefore critical to avoiding the reproduction of (cis)gendered and racist stereotypes. The paper's empirical focus is the EU-funded project iBorderCtrl, designed to manage security risks and enhance the speed of border crossings for third country nationals via the implementation of several AI-based technologies, including facial recognition and deception detection. By drawing together literature from 1) risk and security 2) AI and ethics/migration/asylum and 3) race, gender, (in)security, and AI, this paper explores the implications of lie detection for both regular border crossings and refugee protection with a conceptual focus on the intersections of gender, sexuality, and race. We argue here that AI border technologies such as iBorderCtrl pose a significant risk of both further marginalising and discriminating against LGBT persons, persons of colour, and asylum seekers and reinforcing existing non entree practices and policies.