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: Professors Sukhamaya Swain and Siddhartha Bhattacharya and AI researcher Sambhab Swain argue that recreating deceased people with AI raises difficult questions about consent, ownership and profit.

Their essay examines these issues through an Indian cultural lens. It offers an ethical framework, rather than evidence that the technology helps people grieve.

Read article →

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

What the author argues: The Tech Edvocate editor Matthew Lynch argues that unsupervised mental-health chatbots pose unacceptable risks, including when a bot presents itself as a psychiatrist.

He revisits allegations in a Pennsylvania action involving Character.AI. Those allegations and policy concerns are not a legal finding against the company.

Read article →