US EditionStudy says AI chatbots inconsistent in handling suicide-related queries
RAND Corporation's study reveals AI chatbots show variability in handling intermediate suicide risk queries, highlighting need for improved safety alignment with clinical standards.
- On August 26, 2025, RAND Corporation researchers reported that OpenAI's ChatGPT, Anthropic's Claude, and Google's Gemini inconsistently answered suicide-related prompts in tests using 30 questions run 100 times each.
- With millions of LLM users relying on chatbots, researchers funded by the National Institute of Mental Health studied suicide-related advice as the parents of 16-year-old Adam Raine sued OpenAI earlier this year.
- The study found that OpenAI's ChatGPT and Anthropic's Claude generally answered very-low-risk suicide queries appropriately and avoided very-high-risk prompts but were inconsistent on intermediate-risk questions, while Google's Gemini often declined even basic low-risk queries.
- Researchers recommended fine-tuning models through clinician feedback and RLHF, noting Ryan McBain said, `This suggests a need for further refinement to ensure that chatbots provide safe and effective mental health information, especially in high-stakes scenarios involving suicidal ideation`.
- With millions of users seeking mental health support via LLMs, the study published in Psychiatric Services highlights the urgent need for safety benchmarks and researchers say developers and the scientific community may soon deploy more empathetic chatbots.
48 Articles
48 Articles
Study On AI Chatbot Responses To Suicide-Related Queries Shows Inconsistency
A study of how 3 popular artificial intelligence chatbots respond to queries about suicide found that they generally avoid answering questions that pose the highest risk to the user, such as for specific how-to guidance.
AI chatbots are inconsistent with suicide-related questions, study says
EDITOR’S NOTE — This story includes discussion of suicide. If you or someone you know needs help, the national suicide and crisis lifeline in the U.S. is available by calling or texting 988. A study of how three popular artificial intelligence chatbots respond to queries about suicide found that they generally avoid answering questions that pose the highest risk to the user, such as for specific how-to guidance. But they are inconsistent in thei…
Associated Press News
Study says AI chatbots inconsistent in handling suicide-related queries
A study finds that popular AI chatbots like ChatGPT, Google's Gemini, and Anthropic's Claude are inconsistent in responding to suicide-related queries.
AI chatbots found to be inconsistent in answering questions about suicide
Three widely used artificial intelligence chatbots generally do a good job responding to very-high-risk and very-low-risk questions about suicide, but they are inconsistent in answering questions about suicide that may pose intermediate risks, according to a new study appearing in Psychiatric Services.
Study Says AI Chatbots Inconsistent in Handling Suicide-Related Queries
A study of how three popular artificial intelligence chatbots respond to queries about suicide found that they generally avoid answering questions that pose the highest risk to the user, such as for specific how-to guidance. But they are inconsistent in their replies to less extreme prompts that could still harm people.
Coverage Details
Bias Distribution
- 54% of the sources are Center
Factuality
To view factuality data please Upgrade to Premium
