OpenAI quietly pushed an update to its ChatGPT mobile app this week that turns memory on by default for millions of users.
If you have ever vented to the bot about your boss, your marriage, or your medical anxiety, that context can now persist across conversations unless you dig into settings and switch it off.
The feature, which the company frames as a convenience upgrade, lets the assistant recall your name, your preferences, and details from past chats to make replies feel more personal.
In practice, it means the most intimate confessions Americans make at 2 a.m. are being filed away like grocery lists.
This is the trade we keep making without noticing.
We hand over our inner lives to free apps because the alternative is typing into a void, and then we act surprised when the void starts quoting us back to ourselves.
Privacy advocates have flagged the shift for months, warning that default-on data retention flips the burden onto users.
They will simply keep talking, and the system will keep listening, building a profile that no human assistant would dare keep in a notebook.
This lands as schools, therapists, and even some doctors suggest ChatGPT as a low-cost stand-in for professional help.
A tool that remembers your trauma is not the same as a person trained to handle it, and treating the two as interchangeable is the kind of shortcut that looks efficient right up until it isn't.
There is a broader pattern here that should worry anyone paying attention.
Every few months, another app makes itself a little more essential, a little more intimate, and a little harder to walk away from.
We call it innovation because it is convenient, and we call it convenience because we have stopped asking what it costs.
The fix is not complicated, at least for now.
Open the app, find the memory controls, and decide deliberately what you want a corporation to know about you.
That small act of housekeeping is the difference between using a tool and being used by one.
What makes this moment worth pausing over is not any single update but the direction of travel.
We are training ourselves to treat machines as confidants, and machines are being trained to treat us as data.
Final Thoughts
Somewhere in that exchange, something about privacy, and maybe about dignity, quietly gets traded away.