Keep your own copy of your AI history, privately
The archive runs in your browser with no account and nothing uploaded. What leaves your machine is a list you can read, and it is short.
For anyone who does real work in AI chats and would rather not have a second company holding it.
The situation
Your conversations live on the platform's servers. That is the deal, and for most work it is fine. It stops being fine when the conversation contains client detail, unreleased work, medical questions, or the ordinary half-formed thinking you would not put in a document.
The usual answers all have a catch. Deleting is permanent and unrecoverable, so it protects you by destroying the thing. Temporary chats are not stored at all, which is the point, right up until the answer turns out to be worth keeping. And sharing has gone wrong in public: in 2025, ChatGPT shared links created with the discoverable option enabled began appearing in Google results, and OpenAI removed the option after it was reported.
The tools built to help often make it worse. Saving your AI history into a second product means a second company now holds it, usually on their servers, usually behind an account you had to create.
What Savelore does about it
- It runs offline, with no account
- The extension works entirely in your browser. Entries, folders and tags live in Chrome's local storage on your machine, the way a notes app writes files to your disk. You can use it without signing up for anything.
- Nothing is uploaded unless you switch it on
- Cloud sync is off until you enable it. That is the only moment anything leaves your device, and it is a decision you make rather than the price of using the product.
- What sync sends is a short list
- The entries you chose to save, your folder names, your tags, your preferences and Pro status, and your email for the account. Not your conversation history, not anything you did not save, not your account on any AI platform, and not your browsing.
- Temporary chats, kept without being stored
- You can save an entry from a temporary conversation. The platform still keeps no record of it, and your copy is on your own machine.
- Published pages stay out of search
- If you publish an entry as a public link, that page is readable by anyone holding the link and is not indexed by search engines. Unpublish it and it comes down for everyone. It is the one case where content leaves deliberately, and it covers one entry at a time.
- A business model that does not need your data
- Savelore earns from Pro licences. No ads, no selling data, and no reading your content. There is no version of the product where your archive is the asset.
How to keep a private copy
The default is already the private setup. This is mostly about knowing which switches change that.
- 1
Install it and do not sign in
Saving works with no account. If you never sign in, nothing you save has anywhere to go except your own machine.
- 2
Save the parts worth keeping
A response, a highlighted passage, or a whole conversation. What you do not save is never touched by Savelore at all.
- 3
Decide about cloud sync deliberately
Leave it off and the archive stays on this machine. Turn it on if you need the same archive on another device, knowing it uploads the entries you saved and the list above.
- 4
Export a copy you control
Pro exports to JSON, which is the format to hold if you want a backup that does not depend on Savelore existing.
- 5
Treat publishing as the deliberate exception
Publishing an entry puts that one entry on the open web behind a link. Everything else stays where it was.
What this does not do
Savelore does not make your ChatGPT history private. OpenAI still holds your conversations, and the same is true of Anthropic, Google and every other platform. Saving an entry creates your own copy; it changes nothing on their side.
Deleting an entry in Savelore does not delete the conversation on the platform. If you want something gone from ChatGPT, you have to delete it in ChatGPT, and that is permanent and unrecoverable. Check your platform's own data controls for what it retains and whether your conversations are used for training, because that varies by product and by plan and it changes.
What Savelore controls is the copy you keep. That copy can sit on your machine, under your own labels, without a second company holding it. That is a real thing and it is a smaller thing than making your history private, so it is worth being clear about which one you are getting.
Local-first is a storage decision, not a slogan
Entries are written to your browser's own storage, on your machine. There is no Savelore server involved in saving, which is why the extension keeps working with no network and no account.
The trade is real and worth stating. Local storage lives in your browser profile, so clearing that profile, losing the machine, or reinstalling without exporting first takes the archive with it. Cloud sync and JSON export both exist because of that, and each is a choice you make with the trade in front of you.
Sharing one entry, rather than a conversation
The failure mode people got burned by was sharing a whole conversation on the platform's terms and finding it in search results. Publishing from Savelore sends one entry, as a snapshot, to a page that search engines do not index.
Your other entries, the URL of the source conversation and your email do not travel with it. You can take the page down at any time and it stops loading for everyone, including people who already have the link.
Working somewhere with rules about this
If your employer or client has a policy about where work material can be stored, an archive that never leaves your machine is easier to reconcile with it than one that syncs to a vendor by default. That is a genuine reason people choose local-only mode.
It is not legal advice and it is not a compliance certification. If you are handling regulated material, the question of whether an AI chat should have seen it in the first place comes before the question of where your copy of it lives.
Frequently asked questions
Does Savelore make my ChatGPT history private?
No, and nothing can. OpenAI still holds your conversations. Savelore keeps a copy of what you chose to save on your own machine, which is a different thing.
Can Savelore read my conversations?
No. The extension runs in your browser and nothing is transmitted unless you turn on cloud sync. Even then, only the entries you explicitly saved are uploaded, and the content is not read.
Do I need an account?
Not to save. The extension works fully offline with no signup. An account is needed for cloud sync, for Pro, and to publish an entry, and each of those is separate.
What exactly gets uploaded if I turn on cloud sync?
The entries you saved, your folder names and structure, your tags, your preferences and Pro status, and your email address for the account. Not your conversation history, not anything you did not save, not your platform accounts, and not your browsing.
Are published pages findable in Google?
No. A published page is readable by anyone with the link and is not indexed by search engines. Unpublishing takes it down for everyone.
If I delete an entry, is it gone from ChatGPT too?
No. The two are separate. Deleting in Savelore removes your copy. Deleting in ChatGPT removes theirs, and that one cannot be undone.
What happens to my archive if I clear my browser?
A local-only archive lives in the browser profile, so clearing it takes the archive too. Export to JSON, or turn on cloud sync, if that matters to you.
Your copy can live on your machine.
The conversation stays with the platform either way. What you decide is whether a second company also holds the parts you cared enough to keep. Savelore's answer is no by default, with no account and nothing uploaded.