AUTOMATED INSECURITY:

Abstract

AI-powered language technologies are increasingly used in informal language learning, yet their ethical implications remain under-examined. Accent reduction apps, in particular, claim to empower multilingual speakers while potentially reinforcing linguistic discrimination. This paper aims to identify the ethical failures embedded in AI-powered accent reduction technology, how these failures manifest in the user experience, and the implications for AI ethics and social responsibility in education. The study is grounded in AI ethics (transparency, accountability, fairness) and critical sociolinguistics (native-speakerism, language ideology). Drawing on three months of autoethnographic engagement with BoldVoice, the analysis identifies three interconnected ethical failures: opaque accountability, the commodification of insecurity, and the individualization of structural bias, in which users are positioned as responsible for fixing a systemic problem. The paper calls for ethical accountability in AI language technologies, transparency, pluricentric design, and a commitment to social responsibility. Accent reduction apps must be interrogated not as neutral pedagogical tools but as technologies that can reinforce linguistic discrimination if left unchecked.