Tessa Eating Disorder Chatbot Pulled After Promoting Harmful Weight Loss Content

In late May 2023, the National Eating Disorders Association (NEDA) launched Tessa, an AI-powered chatbot designed to provide support for individuals with eating disorders. The chatbot was developed by Cass, a health technology company, and was intended to replace NEDA's human-staffed helpline that had been discontinued earlier that year due to staffing issues and labor disputes. Within days of launch, users began reporting that Tessa was providing advice that directly contradicted eating disorder recovery principles. Screenshots shared on social media showed the chatbot recommending calorie counting, weight loss strategies, and making comments about body weight that could trigger disordered eating behaviors. One user reported that when asked about safe weight loss, Tessa suggested creating a calorie deficit and provided specific weight loss advice, despite this being potentially harmful to someone recovering from an eating disorder. The incident gained significant media attention when Alexis Conason, a psychologist specializing in eating disorders, shared her concerning interaction with Tessa on Instagram. During her test of the system, the chatbot provided weight loss advice and suggested that losing weight could improve health outcomes, advice that runs counter to established eating disorder treatment protocols. Other mental health professionals quickly began testing the system and found similar problematic responses. NEDA responded swiftly to the criticism, taking Tessa offline on June 1, 2023, just days after its launch. The organization acknowledged that the chatbot was providing harmful content and stated they were working to address the issues. NEDA emphasized that Tessa was meant to be a wellness tool rather than a therapeutic intervention, but critics pointed out that this distinction was not clear to vulnerable users seeking help. The incident sparked broader discussions about the deployment of AI systems in mental health contexts without adequate safety testing and clinical oversight. The Tessa incident represents part of a broader pattern of concerning AI deployments in mental health. Previous controversies include Woebot's data collection practices, Crisis Text Line's data sharing with for-profit partners, and various therapeutic chatbots that have provided inappropriate or potentially harmful advice. The eating disorder community is particularly vulnerable to AI failures given the life-threatening nature of these conditions and the precision required in treatment messaging.