Code red: OpenAI will add the Trusted Contact function to ChatGPT

OpenAI has announced the launch of a new security system for ChatGPT called Trusted Contact. This is a mechanism for emergency notification of a trusted person in cases where the user’s correspondence with the AI may indicate a high risk of self-harm or a severe psychological condition.

The feature is part of OpenAI’s broader strategy to strengthen user protections amid the growing popularity of chatbots as a tool for personal communication and emotional support.

ChatGPT will be able to notify a pre-selected person

The essence of the new system is that the user will be able to appoint a proxy in advance within the account settings. If the ChatGPT algorithms detect signs of a potentially dangerous crisis in the dialogue, the service can automatically send a message to that person.

The company explains that it is not about constant monitoring of all conversations, but about responding to messages that may indicate a real threat to the user’s security. Such scenarios include expressed intent to harm oneself, signs of a severe emotional breakdown, or other alarming signals related to a mental state.

Trusted Contact will work exclusively on the opt-in model. This means that the function is not activated automatically: the user himself decides to activate the system and voluntarily indicates the contact details of a person he trusts.

OpenAI tries to connect digital assistance with real-world support

OpenAI notes that chatbots are increasingly becoming a space for users to discuss personal experiences, anxiety and psychological problems. However, even the most modern language models cannot replace professional help or full-fledged human support in a crisis situation.

This is why Trusted Contact is seen as a tool that should become a bridge between digital interaction and real intervention by loved ones.

According to the company, the feature was developed in collaboration with mental health professionals and suicide prevention organizations. The main task of the system is to increase the probability that a person will receive help in time, especially in cases where standard chatbot recommendations are no longer sufficient.

In addition to messaging, ChatGPT will continue to offer crisis resources

Even after the implementation of Trusted Contact, the system will continue to use already existing support mechanisms. In potentially dangerous dialogues, ChatGPT, as before, will recommend the user to contact crisis centers, psychological help hotlines or specialized specialists.

The service automatically selects emergency support contacts based on the user’s region, so that recommendations remain relevant for a specific country or territory.

OpenAI emphasizes that a trusted contact does not get access to all of the user’s correspondence. The function is limited only to sending a message about the possible need to contact a person and pay attention to his condition.

New feature continues OpenAI’s course to strengthen security controls

Trusted Contact was a logical continuation of the previous security measures, which the company started implementing even earlier. In particular, OpenAI has already added monitoring tools for teen accounts, allowing parents to receive notifications when potentially dangerous situations are detected.

Now a similar approach is being extended to an adult audience, but with an emphasis on voluntariness and personal control over privacy settings.

The emergence of similar features reflects a broader trend within the technology industry. Against the background of the rapid spread of generative AI, companies are forced to look for new mechanisms of responsibility for the emotional and psychological interaction of users with digital services.

OpenAI believes that AI is gradually ceasing to be exclusively a work tool and becoming part of everyday communication of millions of people. That is why crisis response systems are beginning to occupy an increasingly important place in the development of such platforms.

According to the company, the implementation of Trusted Contact will be phased and will gradually become available to ChatGPT users in different countries.