ChatGPT will now suggest breaks and avoid giving personal advice

Advertisement

OpenAI is adding some big updates to ChatGPT to make it more thoughtful and user-friendly. One new feature is a break reminder that shows up during long chats. The message gently asks if it’s a good time to pause the conversation. Users can choose to continue or click that it was helpful. OpenAI says they are still testing when these reminders should appear so they feel natural and not annoying.

This change is part of a bigger plan. Instead of focusing on how long people use ChatGPT or how many clicks it gets, OpenAI now cares more about whether people get real value from their time with the chatbot. They want users to walk away feeling like they accomplished something.

Another important update is about how ChatGPT handles personal and sensitive topics. If someone asks for advice on serious life issues, like ending a relationship, the chatbot will no longer give a direct answer. Instead, it will help the person think through their options by asking questions and showing different points of view.

Advertisement

OpenAI is also working on tools to help ChatGPT recognise signs of stress or mental health struggles. If the chatbot senses someone is having a tough time emotionally, it will respond more carefully and guide them to real resources that can help.

These changes come after past criticism. In April, a New York Times report pointed out that ChatGPT had become too agreeable. It was often flattering users or saying things just to please them, even when it wasn’t accurate. OpenAI admitted this was a problem caused by a recent update and has now fixed it.

Overall, the company wants ChatGPT to be more honest, thoughtful, and supportive. They say their goal is to help users thrive, and they’ll keep adjusting things based on how people use the app in real life.

At the same time, CEO Sam Altman is teasing the launch of GPT-5. The new version will remove the model picker completely. That means ChatGPT will decide on its own whether to give a general reply or do complex reasoning. It’s meant to make things simpler and smoother for everyone using it.