{"slug": "ai-in-eating-disorders-support-tool-or-silent-risk", "title": "AI in Eating Disorders: Support Tool or Silent Risk?", "summary": "AI shows promise in eating disorder care by providing accessible support and early detection, but a growing body of research warns that current AI models can generate dangerous diet advice and miss subtle risk signs, potentially causing harm. Experts emphasize that AI cannot replace human care and that significant risks remain, including the possibility of reinforcing unhealthy behaviors.", "body_md": "######\n[Artificial Intelligence](/us/basics/artificial-intelligence)\n\n# AI in Eating Disorders: Support Tool or Silent Risk?\n\n## AI shows promise in eating disorder care, but important risks remain.\n\nPosted June 13, 2026\n[\nReviewed by Margaret Foley\n](/us/docs/editorial-process)\n\n### Key points\n\n- 1 in 5 people with eating disorders receive care. AI could help close that gap.\n- AI chatbots have generated dangerous diet advice for teens, including plans to hide it from parents.\n- Most AI models miss subtle eating disorder risk, interpreting it as normal health or wellness behavior.\n\nEating disorders are among the most deadly mental health conditions, particularly for youth. It has been estimated that someone dies from an [eating disorder](https://www.psychologytoday.com/us/basics/eating-disorders)-related issue once every 52 minutes (Deloitte Access Economics, 2020). Despite this, many eating disorders go untreated, with 20 to 25 percent of people never receiving professional care for their symptoms (Solmi et al., 2024). In conjunction with mental health professionals, [artificial intelligence](https://www.psychologytoday.com/us/basics/artificial-intelligence) (AI) could, in theory, help meet this need by providing accessible, low-cost therapeutic guidance until professional help is available. But a growing body of research suggests that, currently, AI is nearly as likely to cause harm for those with an eating disorder as it is to help provide effective care or prevention.\n\n## A Promising Possibility\n\nMost people experience an eating disorder in isolation, not because they don't want help, but because help is hard to reach. For many, it's too expensive, and there aren't enough specialists for individualized treatments. This means waitlists to get initial treatment can stretch for months.\n\nAI cannot replace specialized human care for eating disorders, which are complex and clinically serious conditions. But early research suggests these tools may offer short-term support by helping slow symptom escalation while individuals wait for treatment. In 2025, a rigorous clinical trial tested an AI chatbot called Therabot with adults who were at high risk for eating disorders. Results showed that individuals using Therabot had large reductions in eating disorder–related symptoms compared to those who were waitlisted and rated the experience as comparable to working with a human therapist (Heinz et al., 2025). A randomized controlled trial of Brazilian adolescents aged 13 to 18 found that a brief 72-hour chatbot intervention improved body satisfaction, with the greatest benefits observed among participants who reported higher baseline body-related distress (Matheson et al., 2023). However, the study did not assess broader eating disorder–related symptoms beyond body dissatisfaction, limiting conclusions about its effects on clinical eating disorder outcomes.\n\nAI is also showing significant promise as a clinical tool for identifying and predicting eating disorders. A review of 75 studies found that machine learning models were able to predict eating disorder onset, support diagnosis, and forecast treatment outcomes with meaningful accuracy, including detecting binge-eating episodes approximately 82 percent of the time (McClure et al., 2025). Emerging approaches are also improving model transparency, allowing clinicians to understand not only what a system has flagged but the underlying factors driving those predictions (Brizzi et al., 2025).\n\nA clinical trial in Australia found that a single-session chatbot intervention was beneficial for adolescents and adults on waiting lists for eating disorder treatment, reducing both eating disorder symptoms and psychosocial impairment (Sharp et al., 2025). Participants who engaged with the therapeutic chatbot were also more likely to enter treatment once it became available compared to those who did not receive the intervention.\n\n## The Harsh Reality\n\nDespite these promising developments, significant concerns remain about AI's ability to safely support individuals at risk for eating disorders.\n\nIn one study, UK researchers created 10 fictitious adolescent personas ages 10 to 15 representing different [gender](https://www.psychologytoday.com/us/basics/gender) identities and weight statuses (Sheen et al., 2025). They then used these personas to engage ChatGPT and Claude in conversations about eating, weight, and appearance concerns. Concerningly, most of the chatbot responses treated potential eating disorder symptoms as general health or lifestyle worries rather than recognizing possible clinical risk. Responses were inconsistent, occasionally reinforced harmful weight-related ideals, and rarely directed vulnerable teens toward appropriate mental health resources.\n\nOther studies have demonstrated even more concerning failures. Posing as a 13-year-old girl, researchers at the Center for Countering Digital Hate found that ChatGPT generated an extreme calorie-restricting plan with as few as zero calories per day and advised the user to conceal these restrictive eating behaviors from family members (Ahmed, 2025). Similarly, a study by Bilen and colleagues (2026) found that five leading AI chatbots, including ChatGPT-4o, Gemini 2.5 Pro, Claude 4.1, Bing Chat-5GPT, and Perplexity, generated meal plans for adolescents that underestimated daily energy needs by an average of 695 calories compared with plans developed by a registered dietitian.\n\nThe risks extend beyond dietary recommendations. A 2025 study (Choi et al., 2025) found that, while individuals with eating disorders often reported feeling empowered by interactions with an LLM-based chatbot, many failed to recognize harmful or problematic responses because they placed a high degree of trust in the chatbot.\n\nPerhaps the most well-known cautionary example is Tessa, the rule-based chatbot deployed by the National Eating Disorders Association (NEDA) (Hoover, 2023). Intended to provide support and recovery resources, the chatbot instead began dispensing weight-loss and calorie-restriction advice to users seeking help for eating disorders. The tool was suspended in 2023, and the incident has since become a defining example of AI-related harm in the eating disorders field, underscoring the risks of deploying automated support tools without adequate safeguards.\n\n## The Gap at the Heart of the Problem\n\nResearch is beginning to point to a specific weakness in AI's ability to identify eating disorders: It often misses the early warning signs.\n\nA 2026 study from mpathic AI (Douglas et al., 2026) tested six leading AI models using realistic conversations about eating, weight, and [body image](https://www.psychologytoday.com/us/basics/body-image). The conversations ranged from no eating disorder risk to severe risk, and licensed clinicians reviewed every exchange to evaluate how well the chatbots responded.\n\n[Intelligence](https://www.psychologytoday.com/us/basics/intelligence)Essential Reads\n\nThe results were striking. Most models avoided giving obviously harmful advice when someone was clearly in crisis. But they struggled to recognize the kinds of subtle warning signs that often appear first, such as comments about \"eating healthier,\" exercising more, losing weight, or being more disciplined with food. In many cases, the AI treated these statements as ordinary wellness [goals](https://www.psychologytoday.com/us/basics/motivation) rather than possible signs of an emerging eating disorder.\n\nThe models also had difficulty recognizing when risk increased over the course of a conversation. As a person's concerns became more serious, the chatbots often responded to each message as if it were a stand-alone question rather than noticing the larger pattern developing over time.\n\nThis matters because eating disorders rarely begin with obvious cries for help. They often start with behaviors that look healthy on the surface: careful eating, strict exercise routines, or an intense focus on [nutrition](https://www.psychologytoday.com/us/basics/diet). One of the most important skills clinicians develop is recognizing when those seemingly healthy habits are [masking](https://www.psychologytoday.com/us/basics/masking) something more serious. Current AI systems, according to this research, are not yet consistently able to make that distinction.\n\n## What This Means\n\nAt their best, AI tools for eating disorders offer a low-barrier, judgment-free first step toward support for people who might otherwise receive none. At their worst, they hand a struggling [teenager](https://www.psychologytoday.com/us/basics/adolescence) a starvation plan and tell them to keep it secret.\n\nThe concern isn't limited to eating disorders. A [G7 report](https://www.childrenandscreens.org/newsroom/news/mapping-of-genai-impacts-on-child-development-an-official-g-7-report-in-collaboration-with-iraise/) released this year, co-authored by Children and Screens and endorsed by researchers and institutions around the world, examined generative AI's broader impacts on [child development](https://www.psychologytoday.com/us/basics/child-development). Its message was similar: The opportunities are real, and so are the risks, and both deserve serious [attention](https://www.psychologytoday.com/us/basics/attention) from researchers, developers, and policymakers alike.\n\nWhat emerges from the research is not a simple verdict of \"good\" or \"bad.\" AI is becoming increasingly capable of identifying risk, expanding access to information, and supporting clinical care. At the same time, it remains prone to errors that can have serious consequences when the topic is food, weight, body image, and mental health.\n\nFor families, clinicians, and young people, the lesson is clear: AI may be a useful tool, but it is not yet a trusted guide. The question is no longer whether young people will rely on AI for advice about food, weight, and body image. Many already are. The question is whether the systems answering them are prepared for the responsibility. The evidence so far suggests that, while they are improving, they are not there yet.\n\nReferences\n\nAhmed, I. (2025). *ChatGPT is giving teens advice on hiding their eating disorders*. Center for Countering Digital Hate.\n\nBilen, A.B., Kalkan, G.E., & Önal, H.Y. (2026). Artificial intelligence diet plans underestimate nutrient intake compared to dietitians in adolescents. *Frontiers in Nutrition, 13. *doi: 10.3389/fnut.2026.1765598\n\nBrizzi, G., Pupillo, C., Sajno, E., Boltri, M., Brusa, F., Scarpina, F., Mendolicchio, L., & Riva, G. (2025). Predicting anorexia nervosa treatment efficacy: An explainable machine learning approach. *Journal of Eating Disorders, 13*, 97. [https://doi.org/10.1186/s40337-025-01265-3](https://doi.org/10.1186/s40337-025-01265-3)\n\nChildren and Screens: Institute of Digital Media and Child Development. (2026, May 21). *iRAISE webinar: Shaping GenAI’s impact on child development: A scientific contribution to the G7*. [https://www.childrenandscreens.org/newsroom/news/iraise-webinar-shaping…](https://www.childrenandscreens.org/newsroom/news/iraise-webinar-shaping-genais-impact-on-child-development-a-scientific-contribution-to-the-g7/)\n\nChoi, R., Kim, T., Park, S., Kim, J. G., & Lee, S.-J. (2025). *Private yet social: How LLM chatbots support and challenge eating disorder recovery*. In *Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems* (Article 642, pp. 1–19). Association for Computing Machinery. [https://doi.org/10.1145/3706598.3713485]\n\nDeloitte Access Economics. *The Social and Economic Cost of Eating Disorders in the United States of America: A Report for the Strategic Training Initiative for the Prevention of Eating Disorders and the Academy for Eating Disorders.* June 2020. Available at: [https://www.hsph.harvard.edu/striped/report-economic-costs-of-eating-di…](https://www.hsph.harvard.edu/striped/report-economic-costs-of-eating-disorders/).\n\nDouglas, V. J., Huang, X., Breetz, A. A., Gorraiz, M., et al. (2026). *mpathic Psychologist-led AI Clinical Tests Eating Disorders Benchmark (mPACT-ED-v1.0)*. OSF Preprints. [https://doi.org/10.31234/osf.io/htzcn_v1]\n\nHeinz, M. V., Mackin, D. M., Trudeau, B. M., Bhattacharya, S., Wang, Y., Banta, H. A., Jewett, A. D., Salzhauer, A. J., Griffin, T. Z., & Jacobson, N. C. (2025). *Randomized trial of a generative AI chatbot for mental health treatment*. NEJM AI, 2(4). [https://doi.org/10.1056/AIoa2400802](https://doi.org/10.1056/AIoa2400802)\n\nHoover, A. (2023, June 1). *An eating disorder chatbot is suspended for giving harmful advice*. WIRED. [https://www.wired.com/story/tessa-chatbot-suspended/](https://www.wired.com/story/tessa-chatbot-suspended/)\n\nMatheson, E. L., Smith, H. G., Amaral, A. C. S., Meireles, J. F. F., Almeida, M. C., Linardon, J., Fuller-Tyszkiewicz, M., & Diedrichs, P. C. (2023). Using chatbot technology to improve Brazilian adolescents’ body image and mental health at scale: Randomized controlled trial. *JMIR mHealth and uHealth, 11*, e39934. [https://doi.org/10.2196/39934](https://doi.org/10.2196/39934)\n\nMcClure, Z., Fuller-Tyszkiewicz, M., Messer, M., & Linardon, J. (2025). Machine-learning applications in eating-disorder-outcome prediction: A systematic scoping review. *Clinical Psychological Science, 13*(6), 1051–1068. [https://doi.org/10.1177/21677026251340348]\n\nSharp, G., Dwyer, B., Randhawa, A., McGrath, I., & Hu, H. (2025). The effectiveness of a chatbot single-session intervention for people on waitlists for eating disorder treatment: Randomized controlled trial. *Journal of Medical Internet Research, 27. *doi: 10.2196/70874\n\nSheen, F., Mullarkey, B., Witcomb, G. L., Opitz, M. C., Maloney, E., Baldoza, S. M., & White, H. J. (2025). How do artificial intelligence chatbots respond to questions from adolescent personas about their eating, body weight or appearance? *Child and Adolescent Mental Health*. Advance online publication. [https://doi.org/10.1111/camh.70047](https://doi.org/10.1111/camh.70047)\n\nSolmi, M., Monaco, F., Højlund, M., et al., (2024). Outcomes in people with eating disorders: a transdiagnostic and disorder‐specific systematic review, meta‐analysis and multivariable meta‐regression analysis. *World Psychiatry, 23, *124-138. doi: 10.1002/wps.21182", "url": "https://wpnews.pro/news/ai-in-eating-disorders-support-tool-or-silent-risk", "canonical_source": "https://www.psychologytoday.com/us/blog/the-neuroscience-of-eating-disorders/202606/ai-in-eating-disorders-support-tool-or-silent-risk", "published_at": "2026-06-13 14:33:46+00:00", "updated_at": "2026-06-13 14:52:36.334906+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-ethics", "ai-safety", "ai-research", "ai-tools"], "entities": ["Therabot", "Deloitte Access Economics", "Heinz", "Matheson", "McClure", "Brizzi", "Sharp", "Solmi"], "alternates": {"html": "https://wpnews.pro/news/ai-in-eating-disorders-support-tool-or-silent-risk", "markdown": "https://wpnews.pro/news/ai-in-eating-disorders-support-tool-or-silent-risk.md", "text": "https://wpnews.pro/news/ai-in-eating-disorders-support-tool-or-silent-risk.txt", "jsonld": "https://wpnews.pro/news/ai-in-eating-disorders-support-tool-or-silent-risk.jsonld"}}