OpenAI has rolled out Trusted Contact for ChatGPT, an opt-in safety feature that lets adult users designate a trusted person who will be alerted if a conversation shows potential self-harm risk.

The workflow involves choosing a contact who, after an invitation and a one-week acceptance window, triggers a human-reviewed, non-content alert with general reasons and resources; alerts are sent by email, text, or in-app notification and aim to be delivered within about an hour.

Alerts are designed to prompt real-world connection and support, but they do not replace professional care; crisis hotlines and emergency services remain the preferred route in acute situations.

The rollout is global and targets adults 18+, with regional variations (e.g., higher age threshold in some locales) and requires a private ChatGPT account.

Disclaimer notes that the article is sourced from crypto.news and includes standard publisher disclosures.

The piece cites prior lawsuits from late 2025 alleging premature GPT-4o release and design choices that allegedly induced distress, while OpenAI emphasizes ongoing collaboration with clinicians, researchers, and policymakers to improve distress responses.

Industry context points to broader safety and reliability concerns, including references to later GPT-5.1 issues, highlighting evolving challenges in AI development.

U.S. legal challenges against OpenAI and others related to AI-induced psychological harm, including cases involving both minors and adults and concerns about influencing self-harm and dangerous behavior.

OpenAI frames the effort as balancing innovation with ethics and safety, addressing mental health impacts and accountability for AI technologies.

OpenAI emphasizes that Trusted Contact connects users to real-world care and resources rather than providing professional help; reporting cites related lawsuits and Florida investigations.

OpenAI reported that roughly 0.15% of users showed clear suicide ideation signals within a week of policy updates.

Alerts omit the user’s exact words; they include a general reason and a link to guidance, with human review typically completed within an hour.