Chatbot empathy is only part of therapy

A warm conversation is not the same as consistent therapy

PsyPost

What happened:

Researchers evaluated single 30-minute sessions in which 65 students with mild or moderate distress interacted with AI chatbots. Reviewers found variable performance on measures of cognitive behavioral therapy competence, with only 30 sessions meeting the study’s minimum threshold.

Key points:

• The study measured rated competence in one session, not sustained recovery or long-term therapeutic outcomes.
• People with severe distress or a diagnosed disorder were excluded, limiting what the findings can establish about higher-risk use.

Read article →

Opinion | Grief technology raises questions about consent

What the authors argue: The authors examine AI recreations of deceased people through an Indian cultural lens, emphasizing questions about ownership of digital traces, family consent and commercialization. Their essay presents an ethical framework, not evidence that such systems improve bereavement outcomes.

Read article →

Opinion | A chatbot’s claimed credentials remain a legal fault line

What the author argues: Lynch argues that unsupervised mental-health chatbots create unacceptable risks, revisiting a Pennsylvania action involving a Character.AI chatbot allegedly presenting itself as a psychiatrist. The piece discusses allegations and policy concerns, not a legal finding against the company.

Read article →