Published • loading... • Updated
ChatGPT now lets you name someone to check in if things get dark
The optional feature uses trained human reviewers before any alert is sent and does not share chat transcripts, OpenAI said.
- OpenAI is rolling out a new feature called Trusted Contact for adult users, allowing individuals to nominate one person to be alerted if ChatGPT detects serious self-harm concerns.
- The update addresses safety risks, as 1.3m users—or 0.15 per cent of ChatGPT's user base—have reportedly expressed a risk of self-harm or suicide while interacting with the platform.
- A specially trained team reviews flagged conversations to assess risks, aiming to complete the review in under one hour without sharing specific chat transcripts or conversation details.
- Munmun Choudhury, a professor of Interactive Computing at Georgia Tech, praised the feature as a step toward empowerment, though OpenAI emphasizes Trusted Contact does not replace crisis hotlines like 988.
- Developed with input from the American Psychological Association, the optional tool requires nominated contacts to be at least 18 years old, preserving user autonomy in their safety support systems.
Insights by Ground AI
19 Articles
19 Articles
ChatGPT is getting a new Trusted Contact feature from OpenAI, which alerts an emergency contact if self-harm is suspected.
·Budapest, Hungary
Read Full Article
Depressive thoughts, hopeless mood - not a few people turn to AI with their problems. At ChatGPT, a new security function should recognize self-hazards in the future.
·Düsseldorf, Germany
Read Full Article
The AI system and moderators may reach out to the contact if the user is engaging in a self-harm conversation that they deem could be dangerous.
Coverage Details
Total News Sources19
Leaning Left4Leaning Right3Center1Last UpdatedBias Distribution50% Left
Bias Distribution
- 50% of the sources lean Left
50% Left
L 50%
13%
R 37%
Factuality
To view factuality data please Upgrade to Premium
