ChatGPT Adds Trusted Contact Feature for Safety Alerts in Sensitive Situations

ChatGPT has just been upgraded with a new safety feature called Trusted Contact that allows OpenAI to respond to sensitive conversations involving potential self-harm risks. The feature is optional and allows adults to nominate someone they trust, such as a friend, family member, or caregiver, who may be contacted under strictly defined safety conditions.

Trusted Contact adds real-world safety layer

The rollout is designed to connect digital conversations with real-world support when needed most. If ChatGPT’s automated systems detect that a user may be discussing self-harm in a way that signals serious concern, and trained human reviewers confirm the risk, a Trusted Contact may be notified.

The feature is not designed as immediate surveillance. Instead, it works through a staged process that includes automated detection, user prompts, and human review before any external notification is sent. The goal is to avoid premature escalation while still enabling timely intervention in critical cases.

Trusted Contact also extends beyond existing safety tools like localized crisis helplines already available inside ChatGPT. It is intended as an additional support layer rather than a replacement for professional help or emergency services.

How to add a Trusted Contact

Users can add one adult Trusted Contact from within ChatGPT settings. That person receives an invitation and must accept within one week before the feature becomes active. If they decline, users can choose another contact.
If a potential safety issue is detected, ChatGPT first notifies the user and encourages them to reach out to their Trusted Contact directly, often with suggested message prompts. Only after this stage does human review take place.

From there, a small team of trained reviewers evaluates the situation. Only if they determine that there is a serious safety concern is a notification sent to the Trusted Contact through email, SMS, or in-app alerts. The notification is intentionally limited and does not include chat logs or transcripts.

Users retain full control and can change or remove their Trusted Contact at any time. The contact can also opt out independently through support channels.

Expert-backed safety design and safeguards

At a time when AI safety expectations are under global scrutiny, OpenAI says Trusted Contact was developed with input from clinicians, researchers, and mental health organizations, including its Global Physicians Network and external advisory groups.

The system builds on earlier parental safety alerts for teen accounts and expands similar protections to adults. Experts involved in the development highlight social connection as a key protective factor in crisis situations, emphasizing the role of trusted relationships in real-world support.

Alongside this, ChatGPT continues to strengthen broader safety systems. These include encouraging users to seek crisis help when appropriate, improving detection of distress signals, limiting harmful instructions, and suggesting breaks during prolonged use. Not to foeget, OpenAI has also launched a new Codex Chrome plugin and GPT-5.5-Cyber in limited preview.