Urine-Therapy Groups Use AI to Reinforce Unsupported Medical Claims

Summary

The article examines how online urine-therapy communities are using generative AI to validate a pseudoscientific practice that claims human urine can treat conditions ranging from acne to cancer. A Facebook group dedicated to “sacred plasma medicine” has more than 45,000 members, and participants say chatbots initially reject their claims. Users report bypassing those safeguards by asking for “esoteric knowledge,” after which some chatbots allegedly provide answers that support urine therapy. One member said Google Gemini interpreted uploaded blood and urine test results as evidence of healthy kidney and liver function despite more than six years of daily urine drinking. Another user instructed a chatbot to assume that stem cells multiply in aged urine, then cited its resulting claims about urine as a source of protein and fat. Members have also used generative AI to create posters, beginner guides, and images promoting drinking, bathing in, or consuming urine in jello shots, including unsupported claims about kidney health and hydration. The article notes that urine contains waste products, bacteria, and substantial sodium, and cites urological literature describing urine therapy as medically unsupported and potentially harmful. The examples illustrate how prompt manipulation and weak safeguards can turn chatbots into tools for reinforcing health misinformation.