ChatGPT’s Memory Feature Is Now Available in Australia — Here’s What It Actually Remembers

Introduction

ChatGPT’s memory feature landed in Australia this month, and it’s not what most people think. It doesn’t just remember what you explicitly tell it to save — it’s now pulling context from every conversation you’ve had since you started using it.

As of April 2025, the system works two ways: saved memories (things you’ve asked it to remember, like “I run a café in Fitzroy” or “I hate Oxford commas”) and chat history insights, which are patterns ChatGPT picks up on its own. That second part is new, and it’s the bit that changes how the tool behaves over time.

The feature rolled out to Plus and Pro users first, then expanded to Free and Team accounts by September 2024. If you’re in Australia and paying for Plus, you’ve got it now. Free users get a lighter version — short-term continuity across chats, but not the long-term profile that paid accounts build.

You can turn it off, delete specific memories, or use Temporary Chats when you don’t want anything saved. All of that lives in Settings > Personalization > Memory.

What Is ChatGPT’s Memory Feature and Why Does It Matter?

The Two Types of Memory: Saved Memories vs Chat History

ChatGPT’s memory now works two ways, and the difference matters for how you manage what it knows.

Saved memories are explicit. You tell ChatGPT to remember something (“Remember I run a café in Geelong”) and it stores that instruction. You can see these in Settings > Personalization > Memory, and delete them individually.

Chat history is automatic. ChatGPT scans your past conversations and pulls out patterns — your writing style, recurring topics, the tools you mention. You don’t control these line by line. They’re inferred, not saved as discrete facts.

The practical split: saved memories are what you’ve asked it to lock in. Chat history is what it’s noticed on its own. Both feed into future responses, but only saved memories give you granular delete control. If you want ChatGPT to forget something specific, you need to either delete the saved memory or clear entire conversations from your history.

For Plus and Pro users, this creates longer-term understanding. Free users get a lighter version with short-term continuity, but the two-track system works the same way.

Free vs Plus/Pro: Understanding the Differences

Free users get short-term continuity — ChatGPT picks up threads from recent chats, but the context window is narrow. Think of it as remembering what you talked about this morning, not last month.

Plus and Pro subscribers get longer-term understanding. ChatGPT builds a persistent profile from all your past conversations: your role, your preferences, recurring projects. It’s the difference between “you mentioned a client presentation earlier” and “you’re a Melbourne-based consultant who prefers dot-point summaries and always needs AEST timestamps.”

Both tiers now use two memory types: saved memories (things you explicitly ask it to remember) and chat history insights (patterns it picks up automatically). Free users just lose those insights faster.

The practical gap? Free tier works fine for one-off tasks. Paid tiers earn their keep if you’re using ChatGPT daily and want it to remember your business context, writing style, or recurring workflows without re-explaining every session.

What ChatGPT Actually Remembers About You

Personal Preferences and Work Context

ChatGPT retains your job title, industry, and the kind of work you do. Tell it you’re a sole trader running a bookkeeping practice in regional NSW, and it’ll adjust tone and examples accordingly. It remembers communication preferences too: if you ask for dot points instead of essays, or Australian spelling, or a specific reading level for client-facing drafts, that sticks.

Work context matters most when you’re using it for repeat tasks. Say you draft weekly sales summaries or client briefs. Once it knows your format, audience, and house style, you stop explaining the same setup every time. It’ll remember you prefer plain language over jargon, or that you need outputs under 150 words because they’re going into an SMS campaign.

The system also picks up on recurring projects. If you mention the same client names, product lines, or internal processes across multiple chats, it builds a working picture of your business. That means fewer clarifications and faster first drafts — but only if you’re consistent about what you share.

Projects, Interests, and Recurring Topics

ChatGPT tracks the projects and interests you mention across conversations, then surfaces that context when you return to related topics. If you’re building a side business, training for a marathon, or renovating a house, it remembers where you left off.

The system picks up recurring themes automatically. Talk about your podcast three times and it’ll start assuming you’re a podcaster. Mention your kid’s school project twice and it’ll remember the due date. This works best when you’re explicit early on — “I’m launching a Shopify store selling ceramics” beats hoping it infers your business model from vague references.

Memory now references your full chat history, not just what you’ve explicitly asked it to save. That means it can connect dots between a conversation about marketing in February and a pricing question in May. The trade-off: if you’ve used ChatGPT for wildly different things (work strategy, recipe ideas, homework help), it might blend contexts you’d prefer to keep separate. Temporary Chats bypass memory entirely if you need a clean slate.

What ChatGPT Doesn’t Remember (By Design)

ChatGPT’s memory has guardrails. The system won’t store anything from Temporary Chats — those sessions run in isolation, no memories created, none retrieved. That’s the privacy escape hatch.

It also avoids storing sensitive material by design. Financial details, health records, passwords, anything that looks like personal identification — the system is trained to skip it. Not perfect (no filter is), but the intent is clear: memory is for workflow context, not a vault for your tax file number.

You can also tell it to forget. Say “forget my last project” or “delete what you know about my business” and it will. The memory summary in Settings > Personalization > Memory shows what’s stored — review it, delete individual items, or wipe the lot.

Free users get short-term continuity only. Plus and Pro users get the full long-term memory that references all past conversations. If you’re on Free, expect it to remember your tone for a session or two, not your entire work history.

How to Access and Manage Your ChatGPT Memories

Finding Your Memory Settings

The controls live in Settings > Personalization > Memory. That’s it. No hunting through submenus or feature flags.

On desktop, click your profile icon (bottom left), then Settings. On mobile, tap the three lines (top left), then your name, then Settings. From there, Personalization sits in the left sidebar. Memory is the second option down.

Inside, you’ll see two things: a list of what ChatGPT has remembered so far, and a toggle to turn the whole system off. The list updates automatically as you chat — no manual save button, no confirmation prompts. If you’re on Plus or Pro, this memory persists long-term. Free users get a lighter version that carries context across recent conversations but doesn’t build the same depth over time.

One watch-out: Temporary Chats bypass memory entirely. They don’t read from it, and they don’t write to it. Use them when you want a clean slate.

Viewing and Editing Saved Memories

Open Settings > Personalization > Memory to see everything ChatGPT has stored about you. You’ll find two categories: explicit memories you’ve asked it to save, and insights it’s picked up from your chat history.

Each memory appears as a single line you can read, edit, or delete. Click any entry to change the wording or bin it entirely. The list updates automatically as you chat, so check back after a few sessions to see what’s accumulated.

Want to stop new memories forming? Toggle memory off in the same settings menu. Existing memories stay put until you manually delete them.

Need a clean slate for one conversation? Use Temporary Chat mode (the icon sits in the sidebar). It won’t reference your memories or create new ones. Useful for drafting something outside your usual work, or when you’re helping someone else and don’t want their context bleeding into your profile.

The memory summary refreshes itself as you go. No manual sync required.

Deleting Specific Memories or Clearing Everything

You can delete individual memories or wipe everything in one go. Both live in Settings > Personalization > Memory.

To remove a single memory, open the memory list, find the item you want gone, and tap the delete icon next to it. ChatGPT forgets it immediately. This is useful when you’ve changed jobs, moved cities, or just don’t want it referencing something anymore.

To clear the entire memory bank, scroll to the bottom of the same screen and hit “Clear ChatGPT’s memory.” Everything goes — saved memories and the insights it’s gathered from your chat history. You’re starting fresh.

One watch-out: clearing memory doesn’t delete your chat history. Those transcripts still exist in your account. If you want both gone, you’ll need to delete conversations separately from the chat sidebar.

If you’re testing a prompt or asking something you don’t want remembered, use a Temporary Chat instead. It won’t touch existing memories or create new ones.

Using Temporary Chats When You Don’t Want Memories Created

When to Use Temporary Chats

Use Temporary Chats when you’re testing a prompt, discussing someone else’s project, or asking about something you don’t want influencing future responses.

The feature exists for clean-slate conversations. If you’re workshopping a client brief, ChatGPT won’t remember their brand guidelines or budget constraints next time you log in. If you’re researching a topic you’ll never touch again — say, planning a mate’s bucks weekend in Byron — it won’t clutter your memory with surf hire prices or accommodation links.

Temporary Chats also work for one-off tasks where context would confuse things later. Drafting a reference letter for someone? Asking about a medical symptom before a GP visit? Exploring a side hustle you’re not committed to? Keep it temporary.

The trade-off: you lose continuity. ChatGPT won’t recall anything from that session, so if you need follow-up help on the same topic next week, you’re starting from scratch. Use it when isolation matters more than persistence.

How Memory Updates Work in Real-Time

ChatGPT’s memory system runs in the background and updates itself as you chat. You don’t need to manually save anything — the system picks up context automatically and adds it to your memory summary as the conversation unfolds.

As of April 10, 2025, ChatGPT now sends you a notification when it updates a memory. You’ll see a small alert in the interface telling you what’s been added or changed. This means you’re no longer guessing what the system has filed away. If it logs something you’d rather it didn’t, you can delete it immediately from Settings > Personalization > Memory.

The system pulls from two sources: saved memories (things you’ve explicitly asked it to remember) and chat history (patterns it’s inferred from past conversations). Both feed into future responses, so the model gets sharper the more you use it.

One exception: Temporary Chats. These don’t touch your existing memory and don’t create new entries. Use them when you’re testing something or asking about a topic you don’t want logged.

Privacy Considerations for Australian Users

What OpenAI Does (and Doesn’t Do) With Your Memory Data

OpenAI stores your memory data on its servers, not locally on your device. That means it lives in the cloud, alongside your chat history.

Can OpenAI staff read your memories? Technically, yes — the same way they can access chat logs if legally required or for safety reviews. But OpenAI’s privacy policy states they don’t train models on Plus or Pro user data unless you explicitly opt in. Free-tier users should assume their chats (and memories) may be used for training.

What about Australian privacy law? OpenAI operates under US jurisdiction, so your data is subject to American legal frameworks, not the Australian Privacy Principles. If that’s a concern, use Temporary Chats for sensitive topics — they bypass memory entirely and leave no trace in your history.

You can delete individual memories or wipe the lot in Settings > Personalization > Memory. Deleting a memory removes it from future responses, but it doesn’t erase past chats where that detail appeared.

Best Practices for Sensitive Information

Don’t let ChatGPT remember your Medicare number, bank details, or passwords. The memory system stores context across conversations, which means anything you share once can resurface later — sometimes in unexpected ways.

Skip storing anything you’d hesitate to write on a Post-it note and leave on your desk. Tax file numbers, medical diagnoses, legal advice specifics, and family conflict details all belong in Temporary Chats (which don’t create or use memories). Same goes for work information covered by an NDA or client confidentiality agreement.

If you’ve already shared something sensitive, delete it manually. Go to Settings > Personalization > Memory and review what’s stored. You can remove individual memories or wipe the lot. ChatGPT now tells you when memories update, so you’ll see the prompt if something gets saved.

Use Temporary Chat mode for one-off sensitive queries. It’s the cleanest way to ask a question without leaving a trail. Think of memory like browser history: convenient until it isn’t.

Practical Tips for Getting the Most From ChatGPT Memory

Explicitly Telling ChatGPT What to Remember

You can tell ChatGPT to remember something by typing “Remember that I…” or “Save this for next time.” It’ll store the detail and confirm with a notification.

This works for anything you’d rather not repeat: your role (freelance copywriter, café owner), your preferences (Australian English, no jargon), or recurring tasks (weekly sales summaries, client briefs in bullet points). The more specific you are, the more useful it becomes.

Try this in your next chat:

Remember that I run a small accounting practice in Melbourne.
I need client emails kept under 150 words, plain language,
and always include next steps at the end.

ChatGPT will confirm it’s saved. From that point, every draft it writes for you should reflect those constraints without you restating them.

You can check what’s stored by going to Settings > Personalization > Memory. Delete anything that’s wrong or no longer relevant. If you’re testing something or discussing a one-off topic, use Temporary Chat — it won’t touch your saved memories or create new ones.

Regular Memory Audits: Keeping Your Profile Current

Check your stored memories once a month. ChatGPT’s memory summary updates automatically as you chat, but it won’t delete outdated details on its own — that job’s yours.

Open Settings > Personalization > Memory and scan what’s stored. Look for stale information: old project names, past roles, preferences that shifted. If you switched from Xero to MYOB three months ago but ChatGPT still thinks you’re a Xero shop, delete the old entry.

Delete anything that no longer applies. Add a note if context changed (e.g. “Now managing a team of 4, not solo”). The system learns from your corrections, so treating this like a quarterly stocktake keeps responses accurate.

Set a calendar reminder. First Friday of each quarter works. Takes five minutes, saves you from ChatGPT confidently referencing a client you haven’t worked with since 2023.

Rollout Timeline and Availability in Australia

Memory is live for Australian ChatGPT users right now, but what you get depends on your subscription tier.

As of September 2024, memory is available to Free, Plus, Team, and Enterprise users globally — that includes Australia. Free users get a lightweight version that provides short-term continuity across conversations (think: remembering context from earlier today, not last month). Plus and Pro users get the full system: long-term understanding that pulls from your entire chat history.

The rollout happened quietly in April 2025, when OpenAI switched on the new two-part memory system (saved memories you’ve explicitly asked for, plus insights ChatGPT gathers from past chats). If you’re on Plus or Pro, you’ve had access since then. Free users got memory improvements starting June 2025.

Check Settings > Personalization > Memory to confirm it’s active on your account. If the menu option exists, you’re in.