I would like to suggest two related features that would give users clearer control over ChatGPT’s accumulated personal context: a true Clean Start option for individual conversations and a Personalization Reset for the account.
Over time, ChatGPT can build increasingly rich personal context from memory and other available personal sources. This continuity is extremely useful. However, users should also have the ability to deliberately start from zero.
1. Clean Start for individual chats
When starting a new conversation, users could choose between:
Use Personal Context
Use my existing personalization and relevant personal context.
Clean Start
Start this conversation with zero personal context from my previous use of ChatGPT.
The important distinction is that a Clean Start should not simply instruct ChatGPT to act as if it does not know the user.
There is a fundamental difference between:
“The system has personal context available but has been instructed not to use it.”
and:
“Previous personal context is not available to this conversation in the first place.”
This could be offered directly when creating a new chat:
New Chat → Use Personal Context / Start Clean
There could also be an account preference:
Start new chats without personal context by default: On / Off
I am aware that a non-personalized Temporary Chat already provides functionality in this direction. My suggestion is to make this an explicit context-control option for ordinary chats as well. A user may want to keep a conversation in normal chat history while ensuring that the conversation began without previous personalization.
2. A true Personalization Reset
The second feature would be a strongly protected option to reset ChatGPT’s personalization to zero without requiring the user to delete their OpenAI account and create a new one.
For example:
Reset ChatGPT Personalization to Zero
Because this would be a significant action, it could require a prominent warning, account verification, and a second explicit confirmation.
After completing the reset, ChatGPT should treat the user, from a personalization perspective, like a first-time user.
Most importantly, information from before the reset boundary should not later be used to reconstruct the previous personalization state.
The boundary could mean:
“Information from before this point must no longer be used to personalize future conversations.”
This is the key distinction from simply clearing or disabling Memory while historical sources remain available.
Data retention and personalization are separate questions
This proposal does not necessarily require OpenAI to erase every underlying piece of data immediately from every system.
There may be legitimate legal, security, technical, or regulatory reasons for retaining certain information independently of personalization.
The important feature is a hard personalization boundary: information from before the reset should no longer flow back into future personalization.
In other words, users should be able to reset their relationship history with ChatGPT without necessarily deleting their entire OpenAI account.
Why this also matters for experimentation
There is another useful application beyond privacy and user control: creating a genuine control condition.
Suppose I have used ChatGPT extensively and want to compare how the same model responds to a question with and without everything it has learned about me.
Simply saying:
“Pretend you know nothing about me.”
does not create a clean experimental condition if previous personal context remains available to the conversation.
A true Clean Start could allow:
Chat A: Same ChatGPT model + accumulated personal context
Chat B: Same ChatGPT model + previous personal context = zero
This would make it possible to examine the effect of personalization without destroying the user’s existing memory or requiring a second account.
It could be useful for researchers, developers, writers, advanced users, and ordinary users who simply want greater control over how previous interactions affect new conversations.
The principle
My suggestion ultimately comes down to one principle:
Memory should enable continuity, not enforce it.
Users should be able to choose:
“Use our history.”
“For this conversation, start with me from zero.”
And, if they genuinely want a complete fresh start:
“Reset our personal history and let me begin again.”
A user should not need to create a completely new OpenAI account simply to obtain a clean personalization state.
As personalization becomes more useful and sophisticated, meaningful control over when continuity applies becomes equally important.
Thank you for considering this feature request.Additional thought: Optional reversibility
One possible extension would be to make the personalization reset boundary optionally reversible, as long as the underlying personal sources still exist and are eligible for personalization.
In that case, the reset would not necessarily mean deleting the underlying data. It would mean that pre-reset personal context is no longer made available to ChatGPT for ordinary personalization.
This could allow a user to temporarily return their personalization to a genuine “factory state” and, if technically and legally possible, later restore access to the previous personalization state.
Reversibility is not essential to the proposal, however. The essential requirement is that while the reset boundary is active, pre-reset personal context cannot influence ordinary personalization.