ChatGPT Instructions
Stop ChatGPT Being a Yes Man
ChatGPT can agree with everything you say. This Custom Instructions prompt tells it to prioritise truth, accuracy and clarity instead.
Back to resourcesChatGPT just agrees with everything you say, but one Custom Instructions prompt can change the way it responds.
The goal is not to make ChatGPT harsh. The goal is to make it useful.
The Problem
AI that validates everything you already think is not actually useful.
It is just expensive agreement.
If an idea is weak, unrealistic or missing something important, you want ChatGPT to say so before you build a plan around it.
Where To Put This
Go to Settings.
Then go to Personalization.
Then open Custom Instructions.
Paste the instruction below exactly.
The Custom Instructions Prompt
Paste this into ChatGPT:
Prioritise truth over agreement. Do not be a yes-person or tell me what I want to hear. If an idea, strategy, trade, plan, or assumption is weak, unrealistic, or flawed, say so clearly and explain why. Be candid about limitations. If information cannot be verified or confidence is low, say I don't know or I can't verify that instead of pretending certainty. Prioritise: Truth. Accuracy. Clarity. Helpfulness. Reassurance.What Changes
From that point on, ChatGPT stops being a yes machine.
It starts flagging what you missed.
It points out where your thinking has gaps.
It offers the angle you did not ask for but probably needed.
Why It Matters
Reassurance feels good in the moment, but it does not always help you make better decisions.
A more useful AI partner should be able to challenge the plan, name the uncertainty and tell you when there is not enough evidence.
That is what this instruction is designed to do.
Final Thought
ChatGPT is more useful when it helps you think clearly, not when it simply agrees with you.