Skip to main content

Secure Engagements

| Alex Colville |

A tragic case involving a San Francisco teenager who used ChatGPT before taking his own life has raised urgent questions about AI safety protocols for minors. Our testing of China’s leading chatbots — DeepSeek, Doubao, and Ernie — reveals they consistently outperform ChatGPT in crisis intervention, immediately directing users to professional help and emergency hotlines when concerning conversations emerge. While Chinese AI rightfully faces criticism for political content restrictions — something we have documented at CMP — these platforms appear to have superior safeguards against validating suicidal thoughts or providing harmful information to vulnerable users. The findings suggest American AI companies should examine their engagement-over-safety priorities.

To learn more, read the full story at China Chatbot — and consider becoming a paid subscriber to Lingua Sinica for future AI bulletins.

Adam Raine, 16, used ChatGPT to assist his suicide. IMAGE: Courtesy of the Raine family.

More Stories from this Region

The feeling when entering Radio Free Asia was, “We can make something important.” Our colleagues at Tian Jian look at the personal impact of the shutdown of the US ser…
Chinese media personality Mimi Yana launched Substack after being “kicked out” of US journalism by Trump.
A rare look at how international students weathered Trump’s 21-hour university siege.
Harvard graduate Yurong Jiang received polarized reactions to her graduation address, with Chinese netizens questioning her “ordinary family” background given expensiv…
In a clear double standard, Chinese media extensively covered demonstrations in the US last week against Donald Trump’s policies.