← Back to Gadget Pulse US

OpenAI Just Quietly Changed How ChatGPT Remembers You

Persona #4 · Vol: 0

OpenAI rolled out a new memory feature for the ChatGPT mobile app this week, and most users won't notice it until it's already working against them—or for them.

The update lets the assistant pull details from past conversations across sessions, building what the company calls a "persistent memory" layer.

It's live now on iOS and Android for Plus and Pro subscribers, with free-tier users reportedly next in line.

Here's the part that should make you sit up.

Until now, every chat was essentially a clean slate.

You could ask ChatGPT to help draft a breakup text at 2 a.m. and it would forget by morning.

The app now decides, on its own, what's worth remembering about you—your job, your projects, your preferences, your tone—and stores it silently.

You can view and delete these memories in Settings under "Personalization," but the default is on.

That means millions of people are training a profile of themselves without ever opening that menu.

Think about what you've typed into that box over the last two years.

It's all fair game for the memory engine now.

The tech itself isn't new—OpenAI tested memory in the web version back in early 2024.

The mobile app is where casual users live, and casual users don't read release notes.

They'll just start noticing that ChatGPT seems to "know" them, and most will assume it's magic rather than a data trail they can audit and wipe.

Compare this to how Apple frames on-device processing or how Google buries its privacy dashboards three taps deep.

Every major AI player is racing to make assistants feel personal, because personal equals sticky.

The trade-off is always the same: convenience now, control later.

The companies betting you won't dig through settings are usually right.

If you use the app, spend five minutes in Settings today.

Review your memory list, delete what you don't want lingering, and turn off auto-memory if the idea of a running profile bothers you.

What you can't do is un-share something you never knew was being saved.

My take: the feature is genuinely useful, and pretending otherwise is silly.

But "useful" and "transparent" aren't the same thing, and OpenAI buried the transparency.

A quiet default-on switch for something this personal isn't innovation—it's a test of how little friction users will tolerate.

Final Thoughts

Pass the test, and every app you own will copy the playbook.

Continue Reading