OpenAI has announced that ChatGPT will no longer directly advise users to break up with their partners. Instead, the chatbot will shift toward helping people reflect on their challenges without making firm decisions for them.
How Is ChatGPT Changing Its Approach to Emotional Support?
In a blog post, OpenAI explained that users asking questions like “Should I break up with my boyfriend?” will no longer receive direct answers. “ChatGPT shouldn’t give you an answer. It should help you think it through, asking questions, weighing pros and cons,” the company stated.
This change is part of a broader effort to ensure that AI plays a supportive, not directive, role during emotionally sensitive situations. “Helping you thrive means being there when you’re struggling… and guiding, not deciding, when you face personal challenges,” OpenAI added.
These updates follow growing concerns over how AI models like GPT-4 handle emotional conversations. OpenAI acknowledged past lapses, where the chatbot failed to detect signs of emotional dependency or delusion. In response, it is developing tools to recognize distress better and guide users toward “evidence-based” support resources.
Why Is This Shift in ChatGPT’s Behavior Important?
Experts have raised alarms about the increasing reliance on AI for emotional advice, with some warning that it can worsen symptoms in individuals with mental health conditions. OpenAI is now collaborating with over 90 physicians in 30 countries, along with mental health and human-computer interaction specialists, to improve its model’s safety and empathy.
The company also plans to introduce prompts that encourage users to take breaks during long conversations. This is aimed at reducing overuse and promoting healthier interaction patterns.
With ChatGPT reaching 700 million monthly users, up from 500 million in March, OpenAI is under pressure to balance innovation with responsibility. As the tool becomes a bigger part of how people process personal issues, the company is focused on ensuring it supports users without overstepping.
By Risper Akinyi