The increasing integration of social robots into human environments introduces novel risks, particularly the potential for socio-emotional exploitation through social engineering attacks. Traditional cybersecurity measures are often inadequate against threats that leverage human psychological and social vulnerabilities. This paper reviews the challenges posed by socio-emotional exploitation in human-robot interaction (HRI), particularly in the context of healthcare and social aspects. It examines adaptive defense mechanisms, specifically focusing on approaches utilizing real-time physiological signals. Drawing on the foundational work of Pasquali et al. and related literature, this review details the concept of physiologically-informed defense, analyzes its methodological foundations and empirical findings, and discusses its significance, limitations, and profound ethical implications, while also proposing avenues for future research and broader integration of literature.