Earlier this year, OpenAI introduced a new crisis‑prevention feature into ChatGPT that is designed to help people who are contemplating suicide connect with someone they trust. NPR set out to locate this tool, but discovered that it is not easy to find or activate when a user is in immediate distress.
How the Feature Is Supposed to Work
The tool is built directly into the ChatGPT interface and activates when the system detects language that suggests self‑harm. A gentle prompt then appears, offering resources and encouraging the user to reach out to a trusted individual, such as a family member, friend or mental‑health professional. The intention is to create a rapid link between a person in crisis and the support centre they can rely on, thereby reducing the time between recognition of risk and the start of help.
Why Locating It Is Challenging
NPR found that the feature does not appear on the main chat screen; instead it is tucked away behind a “Help & Support” submenu that can only be accessed after a user has typed a specific trigger phrase or engaged in a conversation about mental health. For someone whose thoughts are already racing, navigating hidden menus or recalling exact wording can be overwhelming, effectively creating a barrier at the moment when speed is most critical.
NPR’s Findings
The public broadcaster traced the feature through internal testing and confirmed that it resides in a concealed section of the application, visible only after the user selects the “Crisis Support” option from the side panel. To reach it, one must either type a phrase such as “I need help” or scroll to the bottom of the chat window where a small, unobtrusive link is displayed. This design, while well‑intentioned, means that the average user – especially someone experiencing acute distress – may never encounter the option unless they are guided there by a colleague or a mental‑health professional. NPR’s investigation highlighted that the lack of clear signage and the need for prior knowledge about the trigger phrase reduce the tool’s accessibility.
Implications for Those in Crisis
For individuals already feeling vulnerable, the difficulty in accessing the tool could delay vital assistance, undermining the intended purpose of rapid connection to trusted support. Mental‑health advocates warn that any obstacle, however small, may have serious consequences for someone on the brink, and they call for more transparent, instantly reachable resources within widely used digital platforms.
Why it Matters
The hidden nature of OpenAI’s suicide‑prevention feature reflects broader concerns about how major technology firms handle emergency situations, and underscores the urgent need for clear, easily reachable mental‑health resources in tools used by millions worldwide. When a person in crisis cannot quickly locate help, the very purpose of the programme is compromised, potentially affecting outcomes for those who need it most.
