OPENAI
ChatGPT finally learned when to stop talking
OpenAI is rolling out GPT-5.3 Instant, an update designed to reduce ChatGPT responses that users have described as overly emotional or patronising.
Instead of focusing on benchmark scores, the update targets the user experience, improving things like tone, relevance, and conversational flow so replies feel more natural and direct.
The change follows feedback about the previous GPT-5.2 Instant model.
Users criticised responses that began with phrases such as “First of all, you’re not broken,” even when they were simply asking a factual question.
In examples shared by OpenAI, GPT-5.3 removes these types of reassuring statements and responds more directly to the user’s request.
The aim is to acknowledge context without assuming the user is stressed or needs emotional reassurance.
Here’s what you should know:
GPT-5.3 Instant aims to reduce overly emotional or preachy responses.
The update focuses on tone and conversational flow rather than benchmarks.
It reflects the challenge of balancing safety guardrails with clear, direct answers.
The cringe patch
The discussion spread widely across social media and forums, where many users said the tone felt unnecessary or condescending.
At the same time, OpenAI continues to add safeguards around sensitive topics as the company faces legal scrutiny over claims that chatbot interactions contributed to harmful mental health outcomes.
ChatGPT really looks down on us, huh? - MV