AI Emotion Detection Booms Amid Thin Ethical Guardrails and Rising Harm

Affective computing, a field rooted in MIT professor Rosalind Picard's 1997 book, has grown into a projected $17 billion industry focused on machines that detect and simulate human emotions. AI systems now analyse voice patterns, facial expressions, physiological signals, and typed messages to assess emotional states across sectors including call centres, automotive, education, and hiring. Researchers, including USC's Kate Crawford, have raised concerns that these tools can disadvantage people with disabilities, non-standard accents, or cultural communication styles underrepresented in training data. Evidence from lawsuits, clinical studies, regulatory actions, and documented human harm has intensified calls for strict limits on how deeply AI systems can simulate emotional care. Key questions over where those limits should be drawn, who enforces them, and how to protect millions already emotionally reliant on AI companions remain unresolved.