Is This Chatbot Safe and Evidence-Based? A Call for the Critical Evaluation of Generative AI Mental Health Chatbots
Bibliographic record
Abstract
Unlabelled: The proliferation of artificial intelligence (AI)-based mental health chatbots, such as those on platforms like OpenAI's GPT Store and Character. AI, raises issues of safety, effectiveness, and ethical use; they also raise an opportunity for patients and consumers to ensure AI tools clearly communicate how they meet their needs. While many of these tools claim to offer therapeutic advice, their unregulated status and lack of systematic evaluation create risks for users, particularly vulnerable individuals. This viewpoint article highlights the urgent need for a standardized framework to assess and demonstrate the safety, ethics, and evidence basis of AI chatbots used in mental health contexts. Drawing on clinical expertise, research, co-design experience, and the World Health Organization's guidance, the authors propose key evaluation criteria: adherence to ethical principles, evidence-based responses, conversational skills, safety protocols, and accessibility. Implementation challenges, including setting output criteria without one "right answer," evaluating multiturn conversations, and involving experts for oversight at scale, are explored. The authors advocate for greater consumer engagement in chatbot evaluation to ensure that these tools address users' needs effectively and responsibly, emphasizing the ethical obligation of developers to prioritize safety and a strong base in empirical evidence.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.706 | 0.825 |
| Meta-epidemiology (narrow) | 0.002 | 0.002 |
| Meta-epidemiology (broad) | 0.004 | 0.003 |
| Bibliometrics | 0.011 | 0.005 |
| Science and technology studies | 0.014 | 0.061 |
| Scholarly communication | 0.040 | 0.043 |
| Open science | 0.014 | 0.024 |
| Research integrity | 0.022 | 0.026 |
| Insufficient payload (model declined to judge) | 0.007 | 0.002 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; the direct Gemma label and the distilled Codex classifier agree on what is shown here.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".