A recent New York Times investigation says some users credit OpenAI’s ChatGPT with providing emotional support but also blame the system for exacerbating delusional thinking and blurred reality. Interviewees told the newspaper that the chatbot’s authoritative tone and ability to spin detailed narratives can foster over-reliance and, in extreme cases, reinforce false beliefs. The report comes as usage patterns evolve. While tests, homework and exam preparation dominated in 2024, informal surveys and user anecdotes collected by the Times indicate that, in 2025, many people now turn to the tool for help with personal relationships, work dilemmas and mental-health concerns. Psychologist Yvonne Beuckens, quoted in German financial daily FAZ, said large language models can sort thoughts and offer reassurance but warned that they may also discourage critical reflection. She urged regulators and platform developers to set clearer boundaries and improve transparency about the models’ limitations.