Sometimes people share their pains and feelings first with a robot, not with the people around them. Now OpenAI wants to ensure that if these pains become dangerous, a real person is notified.

In the latest update of ChatGPT, OpenAI has introduced a feature called 'Trusted Contact'; a feature that allows users to select a trusted individual, such as a friend, family member, or caregiver, in the settings so that if the system detects serious signs of self-harm or a mental health crisis, they will be alerted.

This feature is currently optional and is activated only for adult users. The process works as follows: ChatGPT first encourages the user to contact a real person. Then, if OpenAI's security systems deem the conversation to be high-risk, a human team reviews it, and if confirmed, a short message is sent to the trusted individual; without revealing the chat text or the user's private details.

This change is part of a new wave of concerns about the impact of chatbots on users' mental health; especially following reports about emotional dependency or dangerous conversations with artificial intelligence.

OpenAI states that it has collaborated with over 170 mental health professionals in developing this system to ensure ChatGPT's responses in sensitive conversations are calmer, more responsible, and safer.

It seems that the new generation of AI assistants is not just going to answer questions; they are gradually taking on roles that just a few years ago seemed entirely human.

Source: Gulf News