You told an AI assistant your partner's name, your salary range, or a health question you would not bring up at dinner, and then never thought about it again. It answered, the conversation ended, and the information went somewhere. You just do not know where, in what form, or who else can see it.

That is the part that is quietly unsettling about assistants with memory. They are useful precisely because they remember, but "remember" usually means a database entry you never asked to look at and would not know how to find if you tried.

You can look, though. Every major assistant has a settings page for this, they are just scattered in different places under different names. Here is where to check in ChatGPT, Gemini, and Claude, plus how folk keeps the whole thing in one screen instead of three.

Quick answer: Yes, you can see it. In ChatGPT it's Settings > Personalization > Memory > Manage Memories. In Gemini it's Settings > Saved Info, plus Gemini Apps Activity for the chat log behind it. In Claude it's the memory summary under settings. folk keeps all of it in one dashboard view, your "brain," searchable by person or topic, with one command to wipe it clean.

Why does AI memory feel like a black box?

Because for most people it is one. Memory runs quietly in the background while you chat, with no prompt telling you something just got saved. A Malwarebytes survey of over 1,200 people in early 2026 found 90 percent worried about AI using their data without consent, and 88 percent said they hold back from sharing personal information with tools like ChatGPT and Gemini freely. That instinct is reasonable. The problem is most people stop at "hold back" instead of going one step further and just checking what is already stored.

How do you see what ChatGPT remembers about you?

Go to Settings, then Personalization, then Memory. From there, Manage Memories shows the running list of specific facts ChatGPT has saved, things you told it directly or that it flagged as worth keeping, per OpenAI's own memory documentation. You can delete entries one at a time or clear the list entirely, and you can turn the whole feature off if you would rather it never save anything going forward.

The part that trips people up: this is separate from your chat history. ChatGPT can also reference old conversations for context even without a saved memory entry, and deleting a chat does not delete the memories it produced. Those live in the memory list and need to be removed there too.

How do you see what Gemini remembers about you?

Two places, and they cover different things. Saved Info, under settings, holds specific facts you or Gemini explicitly stored, editable and deletable one at a time. Gemini Apps Activity is the broader chat log memory can pull from, where you can review, delete, or set an auto-delete window, according to Google's Gemini Apps Privacy Hub. If you want a conversation to skip memory entirely, a Temporary Chat does that for just that thread.

What about Claude?

Claude keeps a single editable memory summary rather than a running list, and it is visible in your account settings. The practical difference from ChatGPT worth knowing: delete the conversation a memory came from in Claude, and that entry drops out of the summary automatically, since the two are linked. We cover the fuller mechanics, including exactly when Claude got memory and how its single-document approach differs from ChatGPT's two-layer version, in does Claude have memory like ChatGPT.

Does deleting a conversation delete the memory it created?

It depends which assistant, and this is the detail that catches the most people off guard. In ChatGPT, no, saved memories persist after the source chat is gone, you have to clear both separately. In Claude, yes, deleting the conversation removes what it contributed to the summary too. In Gemini, Saved Info and Apps Activity are two different systems, so clearing your chat activity does not automatically clear an explicit saved fact. None of these are wrong, exactly, they just do not behave the way most people assume.

How does folk make memory visible?

folk, the personal AI assistant that lives in your texts (iMessage, Telegram, and WhatsApp), was built so checking your memory does not mean hunting through three different settings menus:

• One view, not a scattered list. Sign in at folk.com and open your dashboard to see your "brain," a visual map of everything folk remembers, grouped by people, topics, and ongoing things, that you can browse and search rather than scroll through as a flat list. See memory for the full picture.
• Delete by asking, or delete in the dashboard. Tell folk in chat to "forget what I said about my old job," or open the brain and remove individual entries directly.
• One wipe clears both layers. Text CLEARMEMORY, or go to dashboard settings > privacy > erase folk's memory, and it clears memory and chat history together in one step, no separate list to remember to also empty. Your account, plan, and connected apps stay put.
• Group memory has the same visibility. In a shared iMessage or Telegram thread, folk's group memory follows the same rule: anyone in the room can see what is remembered, and leaving the group cuts off access to it.

None of this requires trusting folk blindly instead. The isolation and training policies behind it are laid out plainly on privacy, plans, and keys, so you can check the fine print before you decide the transparency is real.

The bottom line

Every assistant with memory has a way to see what it stored, they are just buried under different labels: Manage Memories, Saved Info, a memory summary. Knowing where to look is most of the fix. Check yours today using the settings above, and if switching to something more transparent sounds appealing, folk's dashboard keeps the whole thing in one place from day one, plus import memory if you already have context built up in ChatGPT or Claude worth bringing along, covered step by step in moving your ChatGPT memory to a texting AI assistant.

Get started with folk and check your own brain first, rather than take anyone's word for what it remembers.