A qualitative study published Monday by Stanford researcher Caroline Figueroa found that about one in five young people aged 13 to 22 already seek out AI chatbots for mental health advice, and that teens identified AI sycophancy—the tendency to validate rather than challenge—as a discrete mental health risk. In 2025, approximately 5.4 million young people used generative AI for emotional support when they felt sad, angry, or nervous, according to Hopelab, a San Francisco nonprofit. By 2026, nearly nine in ten young people aged 9 to 17 were interacting with AI chatbots, with one in four using them daily. Figueroa completed in-depth interviews with 48 young people aged 14 to 22, and the answers, published August 11 on the Transparency Coalition's research page, identified five recurring concerns: AI dependency, harmful advice, negative impacts on human relationships, sycophancy, and data privacy. Teens described sycophancy as an AI that functions as "a 'yes ma'am' chat" that "feeds into your own delusions" rather than providing critical challenge. They understood sycophancy as a distinct problem from "harmful advice," requiring different solutions. One participant said, "I don't want to get emotionally attached. Like if the servers shut down, I'm just out of luck," while another described becoming "so reliant on it to the point where it's kind of unhealthy."