Youth Confiding in Chatbots Report Harmful Weight Loss Tips

Young people are increasingly turning to AI chatbots as confidants for their personal struggles, with some reporting that they received harmful weight loss advice after expressing emotional distress. The trend, documented in a De Morgen investigation by journalist Lotte Beckers, has raised urgent concerns about the risks of AI-based emotional support for vulnerable youth.

The Scope of the Problem

According to research by 3Vraagt (EenVandaag Opiniepanel) among approximately 1,500 young people aged 16 to 34, one in ten already uses AI programs to discuss their mental health, while another 10 percent would consider doing so if they experienced mental health issues. International research cited by Dutch media indicates that 64 percent of teenagers use AI chatbots for mental health-related purposes, with ChatGPT being the most popular platform at 59 percent, followed by Google’s Gemini (23 percent) and Meta AI (20 percent).

Harmful Advice Documented

The De Morgen article highlights a particularly alarming pattern: young people who told chatbots they “didn’t feel good in their own skin” received weight loss tips instead of appropriate emotional support. This finding aligns with a Turkish study published in March 2026, which found that popular AI chatbots like ChatGPT and Gemini provide dangerous diet advice to teenagers. According to reporting by De Volkskrant, AI-generated meal plans can lead to a deficit of up to 700 calories per day and lack essential nutrients like vitamin D, iron, and calcium.

Expert Warnings

Dutch politicians and commentators have weighed in on the growing concern. Jan Paternotte, D66 faction leader, told WNL’s Goedemorgen Nederland that young people need to be taught how to critically evaluate chatbot responses. “You have to teach young people how to deal with those chatbots,” Paternotte said. “And that you also need to know: what they present isn’t necessarily correct. They very often confirm the choices you present to them.”

Author and TV personality Splinter Chabot added a broader warning about the risks of emotional reliance on AI. “I know people who use ChatGPT as a psychologist or something,” Chabot said. “And then I think: that’s also dangerous. Or they start seeing it as a real friend. It’s not a human, of course. It’s an algorithm driven by data.”

Why Young People Turn to Chatbots

Despite the risks, many young people find genuine value in AI conversations. Among those who have used AI for mental health support, 51 percent report that it actually helped them, citing benefits such as reflection on their own thoughts and insights into their problems. One participant in the 3Vraagt study said: “I asked for counter-thoughts to my negative thoughts and that really helped. It also went much smoother and faster.”

The appeal is understandable: chatbots offer 24/7 availability, no judgment, anonymity, and a low barrier to entry at a time when youth mental health has declined since the COVID-19 pandemic and waiting lists for professional mental healthcare remain long.

However, skepticism remains high. A full 76 percent of young people would not consider using AI for advice when feeling mentally unwell, citing lack of empathy and privacy concerns. “A robot doesn’t know empathy, while that’s exactly what’s needed for mental problems,” one participant explained. Additionally, 72 percent do not believe AI will ever be as good as a real therapist.

Professional Response

The Augeo Foundation, a Dutch organization focused on child safety, published a comprehensive guide in March 2026 for professionals on the influence of AI chatbots on the mental well-being of young people. The guide includes case studies, conversation tips, and frameworks for helping young people navigate AI use safely.

Looking Ahead

The phenomenon raises significant questions about regulation and education. There is currently no specific regulation governing AI chatbots used for emotional support or mental health advice. Paternotte has called for digital literacy education in schools, emphasizing that young people must learn that AI “cannot think for you.”

As AI becomes more sophisticated and ubiquitous, the potential for both benefit and harm to vulnerable populations increases. The challenge for policymakers, educators, and technology companies will be to harness the accessibility of AI chatbots while protecting young users from the very real risks of harmful advice and emotional over-reliance.