Private AI journal: what Day One, Rosebud, and ChatGPT do with your entries

The Morse teamPublished November 3, 20268 min read

A private AI journal, one where no company or model provider ever reads your entries, is hard to find: as of October 2026, Day One, Rosebud, Reflection, and Mindsera all send the text you ask AI about to an outside model provider, and journaling straight into ChatGPT puts every entry on OpenAI's servers. The apps differ in what happens around that step. Day One keeps most journals end-to-end encrypted until you use AI, Rosebud and Mindsera have zero data retention agreements with some of their model providers, and only some of them say plainly that entries are never used for training.

We read each app's own privacy policy and help pages on October 24, 2026, and every claim below links to the page it came from. OpenAI's help center refused our reader that day, so the ChatGPT facts come from OpenAI's pages as we last read them in October 2026. This page is about who can read your journal. It isn't medical advice, and it says nothing about how to journal.

Is Day One end-to-end encrypted when you use its AI features?

No, not while the AI is working. Day One says end-to-end encryption is "our default method for all new journals, provided the user has set up or entered an encryption key," and that its servers "can store and sync encrypted content across devices but cannot access or read it" (Day One). That same page explains what an AI feature does to it: content "must first be temporarily decrypted on your device before being sent securely to the AI service," so content from end-to-end encrypted journals "is temporarily not end-to-end encrypted while being processed with AI."

The model can be Day One's own or someone else's. The page says the AI services "may be hosted by our own company, Automattic, or may be a third-party service like OpenAI," and names OpenAI as the current third-party provider. Day One deserves credit for spelling this out; most journaling apps don't. It also says AI features need your consent before first use, that it doesn't keep AI requests "on our servers once they have been processed," and that entries aren't used for training "unless you have explicitly given permission." If you want none of it to leave your device, turn AI off under Settings, then AI Features.

Daily Chat, the conversational journal on the Gold plan, works the same way. Day One says it "sends your messages, along with any memories it has built up, to our AI provider," and that reviewing or deleting individual memories "isn't available in the iPhone, iPad, or Mac apps" (Day One).

Is Rosebud journal private?

Private from other people, and from most of Rosebud's staff, but not from its servers or its model providers. Rosebud's privacy policy, last updated September 8, 2026, says entries are stored on its servers in Google's Firestore, which "automatically encrypts all data in transit and at rest" (Rosebud). That protects data on Google's disks, while Rosebud's own systems can still read entries, since they send them to AI providers. The policy doesn't mention end-to-end encryption.

For the AI, the policy names "OpenAI, Anthropic, Google, Amazon Web Services, and Groq." It says data sent to them is "anonymized, containing only the contents of journal entries, without any user-identifying information," and that Rosebud has "BAAs and Zero Data Retention (ZDR) agreements with these providers." We explain what those agreements cover in what is zero data retention. The contents of a journal entry can still name people, places, and your employer, which no anonymizing step removes.

Two lines don't quite match. Rosebud's help center promises "We will never read what you write" (Rosebud help center). The policy says "Authorized individuals access journal content only when necessary to resolve a technical issue, respond to a support request you have made, or investigate a safety or security concern." It also says "AI systems review real interactions" in automated quality evaluations that "are not reviewed by our staff." The policy doesn't say whether entries train models, either way. Deleting your account removes your entries, but "deleted data is retained for up to 30 days in Google Firestore for disaster recovery purposes."

What do Reflection and Mindsera do with your entries?

Both store entries on their own servers and send what you ask about to an AI provider.

Reflection. Reflection's site says entries "are encrypted with AES-256 at rest, and TLS in transit" on Google Cloud, and that they "are never used to train AI models," including on the free plan (Reflection). Its privacy policy, last updated September 4, 2025, says "Currently we use Google Gemini as our AI Partner," and "if you use our AI feature we will send the entry or journal you've queried to our AI Partners servers" (Reflection). To delete your account, you email hello@reflection.app, and the company "will start the process to delete your information in a reasonable time."

Mindsera. Mindsera is the most direct of the four. Its help page says "End-to-end encryption is impossible to use with AI at the moment" because "AI systems need access to the data in the backend to work," and that data sits on "servers located in EU and US depending on the user location" (Mindsera). Its privacy policy, effective September 9, 2026, names OpenAI and Anthropic, says "Your journal content is not used to train or improve AI models," and uses zero data retention "for eligible Claude API requests," adding that it "does not cover every feature or all technical metadata" (Mindsera). Entries are kept until you delete them or your account, "subject to limited backup, security, legal, and dispute-resolution needs."

Is it safe to journal with ChatGPT?

Less safe than an ordinary chat, because ChatGPT's memory is built to keep the personal details a journal is full of. OpenAI's Memory FAQ says saved memories are stored apart from your chats, so deleting the chat where you wrote an entry doesn't delete what ChatGPT remembered from it (OpenAI). On Free, Plus, and Pro, those chats can also train OpenAI's models until you turn the setting off, which we walk through in does ChatGPT train on your data.

AI journaling app privacy, side by side

From each company's own pages, read on October 24, 2026 (ChatGPT: as last read in October 2026).

Every app above except Morse keeps your journal on a server it controls, and every one sends the words you ask about to a model provider in readable form. On Morse, E2EE is the mode that avoids that.

How Morse keeps a journal-style chat on your device

The weak point in every row is the same. For an AI to answer an entry, a model has to read it, and in each app above that model runs on a provider's servers that see the text in the clear, while the journal itself sits on the app's own servers. Morse moves both halves. Your chats live in your browser, and in an E2EE chat the model runs inside a secure enclave that your browser checks before sending a word.

• E2EE, for the entries no one else should read. Your browser verifies the enclave's cryptographic proof that it's genuine, then encrypts the message so only that enclave can read it. Morse's servers relay text they can't read, and the reply is decrypted only in your browser. This is what Day One gives up when its AI runs. The enclaves are run by third-party providers, not by Morse, and the guarantee rests on the enclave hardware and attestation working as designed. E2EE runs open-weight models like GLM, Kimi, and DeepSeek.
• Private, for daily reflection you'd rather no host keep. Private is the default for a new chat. Morse has zero data retention agreements with its inference, search, and other providers, so they keep nothing, and where a provider confirms zero retention on each reply, Morse checks it and refuses a reply that doesn't.
• Anonymous, to think something through with GPT, Claude, or Gemini without the provider knowing it's you. Our proxy removes your account, email, and IP address before the message goes out. The provider still sees the text and may keep it for a limited time to check for abuse, under its own policy, so leave out names you'd rather not hand over.

Whatever the mode, the history stays in your browser, encrypted, and is locked when you sign out. Unlocking it takes two keys, one that never leaves your device and one our server gives you at sign-in, so a copy of your browser data alone can't be read. Our servers don't keep your messages, and Morse doesn't train AI on your content, sell your data, or show ads. Because we don't store chats, we usually have very little to give authorities, which we explain in can AI chats be subpoenaed. Deleting is final: clearing your browser, losing the device, or deleting your account removes your chats, and Morse can't recover them, so save a copy of any entry you want to keep.

Character chats on Morse show crisis lines when a message reads like suicidal thinking, and that check happens on your device.

Which setup fits what you write?

• Entries about health, relationships, or work you'd never want read: start the chat in E2EE. For health questions in particular, see private AI for health questions.
• A daily log with an AI prompt or two: Private mode, the default.
• Wanting a specific model like GPT or Claude: Anonymous mode, with names left out.
• Staying in Day One: keep AI features off for the journals that matter, since that's the only way they stay end-to-end encrypted.
• Journaling in ChatGPT anyway: turn off "Improve the model for everyone," don't rate replies, and check what memory saved.

Try a chat from the Morse home page without an account, or start free. E2EE chats come with Pro, at $19 a month.