OpenAI is updating ChatGPT with new mental health features to make interactions more responsible and user-friendly. The changes come as the company prepares for the expected launch of GPT-5 and a new budget subscription plan. Here is what’s new.
Gentle nudges for healthy use
OpenAI will now show ChatGPT users reminders to take breaks if they’ve been using the chatbot for a while. The goal, the company says, is not to keep people glued to the screen but to make sure they get the help they need and then return to their lives.
“We build ChatGPT to help you thrive in all the ways you want. To make progress, learn something new, or solve a problem, and then get back to your life,” OpenAI said in a statement.
Better answers to tough questions
The chatbot will soon behave differently when asked emotionally sensitive or high-stakes questions, such as those about relationships. Instead of giving a yes-or-no answer, ChatGPT will aim to help users think through their options, weighing pros and cons in a calm, supportive way.
This move is meant to reduce the risk of ChatGPT being seen as a replacement for real-world advice, especially in areas like mental health or personal decisions.
Signs of distress will be handled with care
OpenAI says it is still working on improving how the chatbot identifies signs of emotional or mental distress. The aim is to ensure ChatGPT does not encourage harmful thoughts or dependency. Instead, it will be guided to suggest evidence-based resources or encourage users to seek help from professionals when needed.
To do this, OpenAI has consulted more than 90 doctors from 30 countries and involved researchers in human-computer interaction to fine-tune how the chatbot responds during longer, more personal conversations.
GPT-5 and a new budget plan on the way
Behind the scenes, OpenAI is preparing to roll out GPT-5, its most advanced AI model yet. CEO Sam Altman recently said in a podcast that testing the new model left him feeling “nervous” because of how fast and capable it was. He even compared the experience to the Manhattan Project, suggesting that GPT-5 could be a big turning point for AI.
To go with this release, OpenAI may also introduce a new subscription tier called ‘Go’, a cheaper option than the current $20/month Plus plan. The Go plan was spotted in the app’s code by a user on X (formerly Twitter), though the company has not officially confirmed it.
New tools and features in testing
In addition to mental health updates, OpenAI is quietly testing features like pinning chats and saving favourites on the web version of ChatGPT. These tools could make it easier for users to manage their conversations.
While GPT-5 is not yet here, these upgrades show OpenAI is trying to balance AI’s power with responsibility, something many experts agree is urgently needed.
– Ends