Several outlets report concerns about AI chatbots marketed as “therapists” or capable of providing psychological support. The warning focuses on how some models may have been trained on large amounts of online material, including content that is not designed by licensed professionals and may include dubious advice.
In the coverage, attention is drawn to training data that can span everyday online “self-help” posts, promotional remedies, and even fictional material such as film scripts. Critics argue this raises the risk that a chatbot’s responses could reflect inaccurate, sensational, or unsafe guidance rather than evidence-based mental health approaches. Some reporting emphasizes that people seeking help—particularly those dealing with anxiety or depression—may use these tools for support, which could delay professional care.
While the outlets agree on the general concern about training sources and potential harm, they differ in emphasis. Some highlight the content breadth in training datasets; others underscore the vulnerability of users who rely on chatbots for mental health help. Across accounts, the central issue is the mismatch between marketing and the safeguards needed for health-related use.
