This scoping review of 70 studies found that AI is used mainly to detect and measure health-related stigma, rather than to reduce it. Evidence that AI can reduce stigma remains early, limited, and concentrated in mental health conversational-agent studies.