AI Can Write the Words, But Who Takes the Responsibility?

In my point of view, AI can help with writing blogs and content or other work. But a lot of content and theory that sounds important is really fake, because it has no proof behind it. Humans were already doing this long before AI: people have always written things that only sound important, either to look smart or simply out of foolishness. AI is only a tool, and it makes this easier to do. Companies also want us to see AI as something human, something that has feelings. So I want to ask: what is a feeling? In technical terms, emotions and feelings are chemical reactions, a kind of process happening in the body and brain. Humans are made of atoms, and AI is a process too. Seen this way, we could say that humans are also a kind of AI. I am not trying to define humans here, but this is where the question leads.
AI is good at writing blogs and content, and at other jobs. But I want us to stop and think about how content and its influence affect humans. Can fake or misleading information affect a person's mental health? And who takes responsibility for it? AI companies seem careful about presenting AI as medical care, and the question of responsibility is still unresolved. Providers of AI chatbots face growing scrutiny from advocacy groups and lawmakers over the effects of chatbots on mental health, especially for children. Researchers at Northeastern University found that safeguards have improved around suicide and self-harm, but companies have not extended the same protections to almost every other mental health condition. So companies are acting, but unevenly, and probably under pressure.
Some research shows that anxiety and stress go up with AI use. A survey of 20,847 people in the US found that those who use chatbots for emotional support or other personal reasons were more likely to report symptoms of depression or anxiety. But this shows an association, not cause and effect. A study of high school students in Indonesia found a similar link, but its authors said causality cannot be established because of the study design. A study of health-science students in the UAE suggested that chatbot use is linked to higher depression and anxiety, which likely reflects help-seeking rather than causation. In other words, people who already feel anxious may be the ones who turn to AI.
Other research shows that AI can help. In a randomized trial of 210 adults, a chatbot called Therabot, fine-tuned by mental health experts, reduced depression symptoms by 51% and anxiety symptoms by 31% in participants with those conditions. But this was a chatbot built for therapy and monitored by researchers, not a general chatbot. So what is the main cause? The research does not give one clear answer. It may depend on who is using AI, how they use it, and whether the tool was designed for mental health at all.
My confusion, and my point of view, is this: if someone needs help, they will find help. The real question is what they will find, and who is responsible when it goes wrong.
Sources

Using AI for advice or other personal reasons is linked to depression and anxiety (NBC News)
The Association Between AI Chatbot Use and Anxiety Among High School Adolescents in Yogyakarta
AI-chatbot use and mental well-being among UAE health-sciences students (F1000Research)
Therabot clinical trial summary (2 Minute Medicine, study published in NEJM AI)
For most mental health conditions, AI remains a liability (Northeastern University)
AI Chatbots Face Rising Legal and Legislative Scrutiny (Kelley Drye)