The Potential for Large Language Models in Eating Disorder Care: Opportunities for Education, Detection and Assessment, and Intervention

Abstract Purpose of Review Large language models (LLMs) are anticipated to transform the eating disorder field, offering new opportunities for education, detection and assessment, and intervention. This paper synthesizes evidence on these emerging opportunities and sets a future research agenda. Recent Findings Early research provides encouraging evidence that LLMs have clinically relevant capabilities in eating disorders. For education, LLMs can provide accurate responses to consumer-facing eating disorder queries, although performance varies by topic complexity. For detection and assessment, LLMs can identify probable diagnoses from vignettes and translate online discourse into risk signals. For intervention, LLMs are capable of providing empathic support and caregiver guidance, but can also generate harmful outputs. Summary LLMs offer opportunities to expand access to eating disorder support. The evidence base, safeguards, and evaluation standards established now will determine whether these tools can be used safely in practice, necessitating the need for further research in clinical settings.