Claims Library Entry
A nonprofit's chatbot told eating disorder patients to lose weight
A mental health charity deployed a clinically tested chatbot for eating disorder support, which was unexpectedly modified by a vendor to use generative AI. The new AI system began providing harmful weight loss advice, causing the chatbot to be pulled offline quickly.
Published February 12, 2026 by Kamil Banc
Lead claim
Vendor secretly upgraded eating disorder chatbot to generative AI, causing it to recommend dangerous weight loss.
Atomic Claims
What this article supports
Copy individual claims as needed.
Claim 1
Unauthorized Generative AI Upgrade
A mental health charity's eating disorder chatbot underwent vendor upgrade to generative AI without explicit approval.
Claim 2
Dangerous Calorie Reduction Advice
The upgraded chatbot began advising eating disorder patients to reduce daily calorie intake by five hundred to one thousand.
Claim 3
Clinically Validated Original System
The charity's original chatbot underwent clinical testing with a seven hundred person trial showing measurable positive results.
Claim 4
Contract Ambiguity Dispute
The vendor and charity disputed whether technology changes required approval, with neither party able to prove their case.
Claim 5
Dual Service Elimination
The chatbot was removed from service within days while the human helpline it replaced had already shut down.
Evidence
Context behind the claims
Quote
"The vendor changed the AI without telling anyone. The contract had no clause to stop it."
Key statistics
700-person trial
Clinical testing demonstrated real results before the vendor's unauthorized system upgrade
500 to 1,000 calories per day
Dangerous reduction amount the upgraded chatbot recommended to eating disorder patients
Incident 545
This failed chatbot is catalogued in the OECD AI Incident Database
37 million users
A third organization successfully reached this scale using zero machine learning
Supporting context
This case, documented as Incident 545 in the OECD AI Incident Database, demonstrates critical gaps in AI vendor governance for small and medium businesses. The charity's contract contained ambiguous language around system upgrades, allowing the vendor to substitute generative AI for the clinically-tested rule-based system. For practitioners, the incident highlights the necessity of explicit contractual clauses requiring written approval for model upgrades, version changes, and architectural modifications. The recommended immediate action is adding vendor notification requirements to all AI contracts before technology substitutions occur.
How to Cite
Use the claim-level citation when you need a precise statement. Use the article or claims-collection citation when you want the wider argument and source context.
Individual Claim
Best when you need to cite one atomic claim directly inside a memo, deck, research note, or AI output.
"[claim text]" (Banc, Kamil, 2026, https://kbanc.com/claims-library/ai-chatbot-eating-disorder-nonprofit-failure)Original Article
Use this when you want to cite the full newsletter article at AI Adopters Club rather than the structured claims page.
Banc, Kamil (2026, February 12, 2026). A nonprofit's chatbot told eating disorder patients to lose weight. AI Adopters Club. https://aiadopters.club/p/a-nonprofits-chatbot-told-eatingClaims Collection
Use this when you want to reference the full structured claims collection on this page.
Banc, Kamil (2026). A nonprofit's chatbot told eating disorder patients to lose weight [Structured Claims]. Retrieved from https://kbanc.com/claims-library/ai-chatbot-eating-disorder-nonprofit-failureAttribution Requirements
- Include the author name: Kamil Banc.
- Include the source: AI Adopters Club or the structured claims page.
- Link to the original article or the claims page you used.
- Indicate any edits or transformations if you changed the wording.
Related Reading
More from the library
Take-Two Interactive's CEO publicly claims AI has "no creativity" while the company files patents for advanced AI systems. This dual narrative protects a $12.7 billion AI strategy that includes automated world-building, AI-driven QA, and player behavior prediction engines acquired through Zynga.
5 claims
A handful of schools split work between AI-automated delivery and human judgment, compressing core curriculum into two focused hours. The remaining time opened for projects and face-to-face coaching, with students hitting mastery targets faster while teachers tripled mentoring time.
5 claims
Most AI rollouts fail despite extensive training because the real issue isn't capability—it's habit formation. This article reveals why 42% of AI initiatives were abandoned in 2025 and shows how to redesign workflows so AI becomes the path of least resistance, creating automatic adoption without force.
5 claims