A notable security step from OpenAI: ChatGPT will notify loved ones in times of crisis

This new step, developed to help users find support in the digital environment, enables ChatGPT to respond during mental health crises. The function, known as "Trusted Contact," activates when the artificial intelligence detects serious dangers such as self-harm or suicidal tendencies in its correspondence with the user. When such risks are identified, a loved one designated by the user is immediately informed, and awareness of the situation is rapidly increased.

The application is left entirely to the user's preference. Those aged 18 and over can easily add a loved one or a person from whom they will receive support to their directory via ChatGPT account settings. For the system to work, the added person must also approve the invitation sent to them within seven days. Users can change this person or disable the feature at any time. Likewise, added individuals can opt out of the system as they wish. When a security risk is perceived, ChatGPT first warns the user to contact the person they have designated.

OpenAI emphasizes the importance the system places on privacy. Notifications strictly do not contain chat content; only a short, basic warning message regarding risky situations is sent. When a potential danger signal emerges, this situation is first carefully reviewed by OpenAI's dedicated safety team. After the risk perceived by the artificial intelligence is manually evaluated, a notification is sent to the loved one via email, SMS, or in-app message only when it is truly critical.

The most notable point of this system is that the artificial intelligence does not make decisions alone, and professional oversight is introduced to prevent false alarms. Thus, both privacy violations are avoided, and the user's close circle is mobilized only in emergency situations.

News Source: 12punto