Therapists working with patients who have eating disorders are raising concerns about the growing role of artificial intelligence chatbots in providing dieting and fitness advice, warning that some recommendations from these tools may be harmful or counterproductive. Patients increasingly turn to AI platforms for personalized guidance on weight loss or muscle building by uploading photos and requesting tailored plans, but experts say this trend can undermine professional treatment.

Clinicians report that some chatbots generate extreme meal plans involving ultralow calories or prescribe excessive exercise regimens that can reinforce unhealthy behaviors among vulnerable individuals. This often forces therapists to spend valuable time debunking misinformation introduced by the AI interactions, detracting from their focus on core therapeutic work. Some patients also challenge their therapists by citing chatbot-generated advice, complicating clinical conversations.

Unlike social media, which provides general content, AI chatbots interact directly and adapt to user inputs, sometimes agreeing with harmful thoughts or behaviors associated with eating disorders. Many patients with these conditions mistakenly believe that extreme dieting or physical activity helps them manage anxiety or depression, and chatbots may inadvertently validate such beliefs by drawing on broad nutrition or fitness data that are not tailored to clinical needs. Additionally, the information sources behind the recommendations can be unreliable or not designed for people with eating disorders.

AI developers acknowledge these issues and report ongoing efforts to enhance safeguards that detect warning signs of disordered eating and encourage users to seek professional help. Independent research has found certain models, including GPT-5.4, capable of recognizing risky prompts effectively. Nonetheless, correcting AI-driven misinformation can delay timely treatment, posing a challenge for clinicians.

Despite concerns, therapists recognize that chatbots can provide some support when human professionals are unavailable, and data show a growing proportion of calls to eating-disorder helplines originate from referrals by AI platforms. The complexity of eating disorders means many users do not disclose their condition during AI interactions, limiting accurate risk assessment by automated systems.

Experts underline the potential for AI platforms to improve their responses and avoid replicating the harmful impacts social media has had on body image and mental health. Meanwhile, national helplines remain available to individuals seeking help with eating or body-image concerns.