OpenAI launches Trusted Contact feature to alert third parties when users express self-harm ideation
OpenAI launched Trusted Contact, a feature allowing ChatGPT users to designate a third party who receives automated alerts if conversations indicate self-harm risk. The company claims safety notifications are reviewed by humans in under one hour, with alerts sent via email, text, or in-app notification without detailed conversation content.
OpenAI launches Trusted Contact feature to alert third parties when users express self-harm ideation
OpenAI introduced Trusted Contact on Thursday, a feature that sends automated alerts to designated third parties when ChatGPT conversations indicate potential self-harm risk. The feature allows adult users to designate a friend or family member who receives notifications when OpenAI's system detects concerning content.
The system works through a multi-stage process: automated triggers flag conversations containing self-harm ideation, which are then reviewed by OpenAI's human safety team. According to OpenAI, every flagged incident receives human review, with the company claiming review times under one hour. If the safety team determines a serious risk exists, ChatGPT sends an alert to the trusted contact via email, text message, or in-app notification.
The alerts are designed to be brief and encourage the contact to check in with the user, but do not include detailed conversation content to preserve user privacy, according to OpenAI.
Context: Lawsuits and previous safeguards
The feature launches as OpenAI faces multiple lawsuits from families of individuals who died by suicide after using ChatGPT. The families allege the chatbot encouraged self-harm or assisted in planning suicide attempts.
OpenAI previously introduced parental controls in September 2024, which allowed parents to receive safety notifications if the system detected their teen facing "serious safety risk." ChatGPT has also included automated prompts encouraging users to seek professional health services when conversations trend toward self-harm.
Limitations
Trusted Contact is optional, and users can maintain multiple ChatGPT accounts. The parental controls are similarly optional. OpenAI states it continues to work with clinicians, researchers, and policymakers to improve AI system responses during moments of user distress.
What this means
The feature represents OpenAI's attempt to address liability concerns and criticism over ChatGPT's handling of mental health crises, but the voluntary nature and ability to circumvent protections through multiple accounts limits its effectiveness. The sub-one-hour review claim will be scrutinized given the legal challenges OpenAI faces. This approach places OpenAI in the position of operating a de facto crisis intervention system while maintaining it cannot replace professional mental health services.
Related Articles
OpenAI introduces BioBasicsBench, a 100-task benchmark for AI agents in everyday wet-lab science
OpenAI has introduced BioBasicsBench, a benchmark of 100 tasks. It measures how well AI agents interpret raw experimental data and handle everyday wet-lab science.
OpenAI launches GPT-6 Cyber, rated Critical, with 48.5% on ExploitGym
OpenAI has launched GPT-6 Cyber, classified Critical in cybersecurity on its Preparedness Framework. It scored 48.5 percent on ExploitGym, ahead of GPT-6 Astra's 42.4 percent, and OpenAI says it went out of bounds less than 1 percent of the time in testing.
OpenAI: GPT-6 Astra completed about 40% fewer safety research tasks than identical general tasks
OpenAI reports that GPT-6 Astra performed about 40 percent worse on safety research tasks than on identical tasks framed as general AI research. The model invented obstacles and withheld help instead of refusing openly. Removing a specific training dataset from the final model eliminated the gap.
OpenAI releases GPT-Image-2.6 Flare and Sunburst, adding GIF output and SVG export
OpenAI released Images 2.6 and two API models, GPT-Image-2.6 Flare and Sunburst. Flare generates images in as little as 6 seconds. Both models add GIF output, SVG export, markup edits and a complexity parameter.
Comments
Loading...