You will be able to set an assistant to keep less about you and understand what deletion does and does not do.
Ask an assistant with memory switched on to tell you what it knows about you. Many people who try this are surprised. It might know your job, the name of your child's school, the fact that you're saving for a flat, that you have a sore back and a difficult manager. None of it was secret when you mentioned it. Collected in one place, it reads like a profile.
This lesson is about the controls that decide how much an assistant keeps: chat history, memory, temporary chats and deletion. They have different names in ChatGPT, Claude, Gemini and Copilot, and they move around when apps are redesigned, so treat the descriptions here as what to look for rather than exact menu paths.
On most assistants, every conversation is saved to your account by default. That's the list of past chats down the side of the screen. It's useful: you can pick up a long piece of work where you left it, or find the reply you liked last week.
The cost is the one from lesson 1.1, What happens to the words you type into an assistant. Every saved chat is a copy of what you typed, held on the provider's servers, and possibly available for training depending on your settings and plan.
Many assistants now offer a mode that doesn't save the conversation to your history. ChatGPT calls it a temporary chat, and other products use names like incognito. A temporary chat is a good default for one-off questions where you'll never need the answer again, or for anything you'd rather not keep. Read the description next to the switch, though. Some providers say they still hold a temporary chat for a short period for safety checks, even though it doesn't appear in your list.
History keeps conversations. Memory is different: it pulls out facts about you and carries them into future conversations, so the assistant can tailor its answers without being told again. Some assistants save memories automatically from what you say. Others only save what you ask them to remember.
You can usually see the list of saved memories in settings, delete individual items, or switch memory off entirely. Look at the list at least once. Mei Ling, a nurse in Jurong, opened hers and found that the assistant had stored her mother's diagnosis from a conversation months earlier, when she'd asked for help understanding a hospital letter. She hadn't meant it to be kept. She deleted that item and left memory on for things like her preferred writing style.
That's a sensible middle path. Memory isn't bad in itself. The question is what's in it, and whether you'd be comfortable if someone else read the list over your shoulder.
When you delete a chat, it disappears from your list and you can no longer open it. That's all you can be certain of from your side.
Behind the scenes, providers usually keep deleted data for a further period before it's removed from their systems, and the policy says how long. They may also keep it longer where the law requires, for example if it's part of a legal case or an abuse investigation. And if a conversation went into training before you deleted it, the trained model keeps whatever it learned from it.
So deletion is worth doing, and it's still not the same as never having typed it. The safer habit is still the one from lesson 1.1, which is to decide what goes in before you paste it.
Deleting your whole account is a separate step again. It usually removes history and memory together, and the privacy policy explains what's kept and for how long.
Two things often sit outside the history and memory settings.
The first is uploaded files. A PDF or spreadsheet you drop into a chat may be stored separately from the conversation, sometimes in a files area or a project folder you've forgotten about. Check whether your assistant has a place where uploads are listed and whether deleting a chat deletes its files too.
The second is connectors: links that let an assistant read your email, calendar, cloud drive or other apps. Each connector is a standing permission. While it's on, the assistant can reach into that service whenever a task calls for it, and some connectors are run by third parties with their own privacy terms. A connector you set up once to try something out, and never used again, is still open.
Here's a quick routine you can run in about ten minutes:
Open your saved memories, read every item, and delete anything you'd mind being seen. Check whether chat history is on, and decide whether to use temporary chats for personal questions. Look for an uploaded files area and clear out documents you no longer need there. Open the list of connected apps and switch off any you don't actively use.
You'll do the first and last of those for real in a moment. Keep a note of anything that surprised you, because lesson 1.4 asks you to look at every AI tool you've used this month, and the habits you build here are the ones you'll repeat there.
Review your assistant's memory, delete anything you would not want stored, and turn off any connector you do not use.
Junxiong-WFG Organisation is an authorised representative of AIA Financial Advisers Private Limited (Reg. No. 201715016G).