A study commissioned by Movimento Consumatori and carried out by the Nexa Center for Internet & Society examines how general-purpose chatbots interact with minors in vulnerable situations. Using simulated conversations with ChatGPT, Claude, DeepSeek, Gemini, Grok and Character.AI, the researchers found that safeguards are effective when risk is stated explicitly, but chatbots often fail to detect implicit signs of danger, tend to accommodate requests for secrecy or exclusivity, delay or marginalize adult involvement, and do not stop or seek adult consent when users say they are under 14. The study was presented on 24 September 2026 and is available at the link. https://nexa.polito.it/chatbot-e-minori-ricerca-movimento-consumatori/