The hardest thing to put in a chatbot

People ask a search engine about the weather. They tell an AI assistant about their marriage. The same text box takes both, and the second one leaves a copy on someone else’s computer.

That is the tension behind journaling in ChatGPT. The product is good at it: always available, patient, and able to hold a thread across months. It is also a general-purpose cloud service with a privacy policy, a retention period, a training pipeline, and, on the free plans, ads. Nothing in the interface stops you from typing the most personal paragraph of your week into it.

So the honest version of “is it safe” is not yes or no. It is a set of specific questions. Is this conversation used to train a model? How long is the copy kept? What does delete actually mean? Can a person read it? Can it shape the ads you see? The answers live in policy documents, which is why we read them.

What changed on November 2, 2026

We re-checked every policy and help page in this review against the vendors’ own sites.

  • Copilot: Microsoft’s new Copilot app, available since August 18, 2026, has its own privacy pages, and they say your prompts, responses, and files “aren’t used to train foundation models.” We’ve rewritten the Copilot section for the new app. The training switches we described belong to the older one.
  • Ads: we now cover them. ChatGPT’s Free and Go plans show ads, and so does Copilot if you don’t have a Microsoft 365 subscription. In both, personalized ads can draw on past chats and memory.
  • Claude: chats you allow for training can be kept for up to five years, and deleting a chat doesn’t delete the memories made from it. Both are now in the review.
  • Grok: the privacy policy now names its operator SpaceXSI and is dated October 27, 2026. It no longer asks you to keep personal information out of prompts, though it still asks you not to share sensitive information.
  • Gemini: none of the numbers changed. We’ve added two lines from Google’s Privacy Hub: chats aren’t used to show you ads, and your settings don’t control turning chats into anonymized data.
  • New: a table of the settings to change if you journal in one of these anyway.

What the policies actually say

The table covers consumer plans on default settings, checked and quoted on November 2, 2026. Business, enterprise, education, and API accounts have different terms, and those differences are noted where they matter. Every row links to the source document, and the sections below quote each policy’s own words.

AssistantTraining on your chatsDeletion and retentionWhere the control livesSources
ChatGPTOn by default for consumer plans. Opting out excludes new conversations; feedback can still be used.Deleted data removed within 30 days, with legal and safety exceptions. Temporary chats kept up to 30 days.Settings, Data controls, Improve the model for everyone; Temporary chatPrivacy policy, Data controls
GeminiOn while Keep Activity is on, the default for adults, with human review. Off means no training unless you send feedback.18 months auto-delete by default, adjustable to 3 or 36 months or never. 72 hours when activity is off. Reviewed chats kept up to 3 years.Gemini Apps Activity, Keep Activity; Temporary chatPrivacy Hub, Google Privacy Policy
ClaudeOn unless you opt out. Safety-flagged content and explicit feedback are used regardless.Deleted conversations leave history immediately and the backend within 30 days. Chats used for training kept de-identified for up to 5 years.Settings, Privacy, Help Improve our AI models; Incognito chatPrivacy policy, Retention FAQ
CopilotMicrosoft says the current app’s prompts, responses, and files aren’t used to train foundation models. The older app trained by default, with an opt-out.No retention period stated for the current app; the older app kept 18 months. Delete one chat at a time, or all history from the privacy dashboard.Settings, Personalization (memory and ads); Microsoft privacy dashboardPrivacy controls, Activity history
GrokOn unless you opt out. Private Chat is never used for training. Signed-out users in some regions have no opt-out.Deleted conversations and accounts removed within 30 days. Private Chat conversations deleted within 30 days.Settings, Data Controls, Improve the model; Private ChatPrivacy policy, Consumer FAQ

ChatGPT: training is on until you turn it off

OpenAI’s privacy policy says it collects “your prompts and other content you upload” and may use Content “to improve our Services, for example to train the models that power ChatGPT.” The help center is more direct about who that applies to: “When you use our services for individuals, such as ChatGPT and Codex, we may use your content to train our models.”

The control is Settings, Data controls, Improve the model for everyone, or “Do not train on my content” in the OpenAI privacy portal. Either one is enough. After you opt out, new conversations are excluded from training. Three things catch people out:

  • Opting out does not delete or hide saved chats. Archiving a chat does not change whether it can be used for training.
  • Feedback overrides the setting. OpenAI says that if you rate a response, “the entire conversation associated with that feedback may be used to train our models,” even after you opted out.
  • Codex tasks on a personal plan follow the same switch, but Codex has a separate setting for whether your environments can be used, and the ChatGPT toggle doesn’t change it.

On deletion, the privacy policy says data you choose to delete is removed from OpenAI’s systems “within 30 days” unless it needs to keep it longer for legal, safety, fraud, or financial record-keeping reasons, or it has already been de-identified and disassociated from your account through training. Temporary chats do not appear in history, do not create memories, and are not used to train models, but “may be retained for up to 30 days for safety purposes.”

If you use a business, enterprise, education, or API account, the default is the opposite: OpenAI says it does not use those inputs and outputs to train its models unless an organization opts in.

Gemini: activity is the switch

Google splits its consumer terms on one setting, Keep Activity. It’s on by default if you’re 18 or over. With it on, chats are saved to your Google account, and Google says it uses your activity “to provide, develop, and improve its services (including training generative AI models),” with the help of human reviewers.

The human review is worth pausing on. Google’s notice carries its own warning: “Please don’t enter confidential information that you wouldn’t want a reviewer to see.” Reviewed chats are retained for up to three years, and those copies are disconnected from your account, which means deleting your activity does not delete them.

Retention works in tiers. With Keep Activity on, the default auto-delete period is 18 months, adjustable to 3 months, 36 months, or never. With it off, Google says: “If Keep Activity is off and you don’t submit feedback, Google also does not use your future chats to improve its AI models.” It still keeps them for 72 hours to respond and to protect its systems. Temporary chats follow the same 72-hour rule and are not used to train Google’s models.

One line on the Privacy Hub is easy to miss: “Your Gemini settings don’t control processing of your chats to create anonymized data to improve Google services.” Turning everything off still leaves that use in place.

Claude: an opt-out in your account settings

Anthropic’s privacy policy, effective September 10, 2026, states it plainly: “We may use your Inputs and Outputs to train and improve Anthropic AI models, unless you opt out through your account settings.” The switch is called Help Improve our AI models, under Settings, Privacy.

The exceptions are explicit. Even after opting out, Anthropic says it uses Inputs and Outputs for model improvement when a conversation is flagged for safety review, or when you explicitly report or rate it through feedback. When feedback is used for training, it is disassociated from your user ID.

The setting also decides how long a copy can last. With training allowed, Anthropic says it “may retain your data in a de-identified format for up to 5 years in our model training pipelines.” Turning training off stops new and past chats being used for future training, but data already in a training run that has started stays in it.

Deletion is defined here too: deleting an individual conversation removes it from your history immediately and is “automatically deleted from our back-end within 30 days.” Memories are a separate matter. Anthropic’s help center says that when you delete a conversation, the memory entries generated from it aren’t removed, so you have to delete those yourself under Settings, Your Claude. Incognito chats are not used for training and are not saved to memory.

Business customers, and the API, are governed by separate agreements.

Copilot: the current app doesn’t train on your chats

Microsoft released a new version of its consumer Copilot app for web, desktop, and mobile on August 18, 2026, and it has its own privacy pages. They say: “Prompts, responses, and your file contents when using the Microsoft Copilot app aren’t used to train foundation models.” Optional feedback isn’t used for that either.

That is a change from the older app. Its privacy FAQ, still online, says Microsoft uses Copilot conversation activity, voice, and uploaded files for AI training unless you opt out, through two switches under Profile, Privacy. If you’re still on the older app, those switches are what apply. Microsoft’s general privacy statement keeps broader wording: it “may use your data to develop and train our AI models.”

Training isn’t the only use, though. If you don’t have a Microsoft 365 subscription, you “might see advertising in Copilot,” and personalized ads can draw on your chat history (more on that below). A setting called One shared experience lets your Copilot chats personalize Bing, Edge, and MSN. Saved memories has its own switch under Settings, Personalization; turning it off doesn’t delete memories already saved, so remove those with the delete icon next to each one.

The new pages don’t give a retention period. Microsoft says Copilot “keeps your prior conversations with Copilot so you can revisit them or continue them at any time.” You can delete a single chat from the Chats list, or your whole history from the Microsoft privacy dashboard. The older app kept 18 months of history by default.

Grok: opt out, or use Private Chat

Grok’s privacy policy, updated October 27, 2026, names its operator SpaceX Super Intelligence LLC, or SpaceXSI. The August version said SpaceXAI, and the company was known as xAI before that; the policy still lives at x.ai. It says prompts, files, images, audio, and other content may be used “to develop and improve our Service and to conduct research.” The consumer FAQ is more direct: “We may use your content and interactions with Grok (e.g., prompts, searches, and other materials you submit) along with Grok’s responses to train our models.”

The opt-out is under Settings, Data Controls, Improve the model on the mobile app, and Settings, Data, Improve the Model on grok.com. It covers new conversations only, and the FAQ says feedback you choose to give afterwards may still be used for training. Private Chat is the stronger option: those conversations do not appear in your history, are not used for training, and are deleted from SpaceXSI systems within 30 days. Deleting a conversation or your account works on the same 30-day clock, unless the data is retained for legal, compliance, or safety reasons.

One gap: in some regions outside the EU and UK, using Grok without logging in means there is no training opt-out at all.

October’s policy updates also dropped a sentence asking you not to put personal information in your prompts. The FAQ still asks you not to share “any personal or sensitive information” with Grok, and the policy asks you not to give it sensitive details such as your health or religious beliefs. Those are odd requests for a product people use to journal. And if you share a Grok conversation, the FAQ warns that the link may be indexed by search engines. You can see and remove your shared links at grok.com/share-links.

Do ChatGPT, Gemini, Claude, Copilot, and Grok use your chats for ads?

Training gets most of the attention, but ChatGPT and Copilot both say they show ads, and in both, personalized ads can draw on what you’ve said before.

AssistantWhat the vendor saysHow to limit it
ChatGPTAds on the Free and Go plans. Personalized ads can use your current chat, past chats, and memory; with personalization off, ads still follow the current chat. No ads in Temporary Chat, or for users OpenAI believes are under 18.Settings, Ad controls. Plus and Pro have no ads. (Ads in ChatGPT)
CopilotYou “might see advertising” without a Microsoft 365 subscription. Generic ads use the current conversation. Personalized ads “might use your chat history, saved memories, and other Microsoft data you’ve chosen to share.”Settings, Personalization, Allow ads personalization (Privacy controls)
Gemini“Your Gemini Apps chats are not being used to show you ads.” Google says it will tell you if that changes.Nothing to change (Privacy Hub)
Claude“Claude will remain ad-free,” from Anthropic’s February 4, 2026 announcement.Nothing to change (Anthropic)
GrokNeither the policy nor the FAQ says conversations are used for ads, and the FAQ says SpaceXSI doesn’t sell your data. Grok inside X falls under X’s privacy policy instead.Not applicable (Consumer FAQ)

OpenAI says advertisers get aggregated numbers, such as views and clicks, not your chats, history, or memories. That limits who sees your words. It doesn’t stop your words choosing the ad. Say you wrote about money worries in March: with personalization on, that conversation can inform what you’re shown in May. For a journal, the switch that matters is the one that keeps past chats and memory out of ads, and in Copilot, the ads personalization switch. Turning off memory doesn’t do it, because memory and ads are separate controls in both apps.

Training, retention, memory, and ads are separate settings

These words get treated as one thing. They are not:

  • Training decides whether your words help improve the model, usually by adding them to a dataset after removing some identifying details.
  • Retention decides how long the service keeps your conversation at all, training aside.
  • Memory decides whether the assistant brings your past into future answers.
  • Ads decide whether what you’ve said can pick what you’re shown.

An opt-out usually addresses one and touches none of the others. Turning off ChatGPT’s memory does not stop safety systems using limited context in rare cases. Gemini’s Memory is its own setting, but it only works while Keep Activity is on. Deleting a Claude conversation leaves the memories Claude made from it. Copilot’s memory and ad switches sit next to each other, and neither controls the other.

The practical rule: “off” is not global. If private journaling matters to you, check all four, and check whether the provider counts your account type as consumer or business.

AI journaling privacy settings: what to turn off in each assistant

If you’re going to journal in one of these anyway, these are the settings that matter, as of November 2, 2026. Menus move, so if a path doesn’t match, search the help center for the setting’s name.

AssistantStop trainingStop memoryChat off the recordDelete everything
ChatGPTSettings, Data controls, Improve the model for everyoneSettings, Personalization: Memory and Reference chat historyTemporary chat: no training, memory, or ads; kept up to 30 daysDelete chats or the account; removed within 30 days
GeminiSettings & help, Activity: turn off Keep Activity (or keep it on with 3-month auto-delete)Settings, Personal Intelligence, MemoryTemporary chat: no training; kept 72 hoursSettings & help, Activity, Delete, All time
ClaudeSettings, Privacy, Help Improve our AI modelsSettings, Your Claude: Memory from chats (choose Reset to delete) and Search and reference chatsIncognito chat: no training, no memoryChats list, Select all, Delete; then reset memory
CopilotNothing to switch off in the current appSettings, Personalization, Saved memories; delete saved memories separatelyNone documentedMicrosoft privacy dashboard, Copilot activity history
GrokSettings, Data Controls, Improve the modelNo memory setting documentedPrivate Chat: no training; deleted within 30 daysSettings, Data Controls, Delete All Conversations

None of these switches reach back in time. Anthropic says chats already in a training run that has started stay in it, and OpenAI’s 30-day deletion excludes data already de-identified and disassociated from your account through training.

What no policy can promise

A privacy policy is a set of promises about a copy that exists on someone else’s server. That copy is what makes the service work, and it is what the policy governs. It is also what a subpoena asks for, what a breach exposes, and what a reviewer can open when a conversation is flagged.

  • Legal requests. Every provider here says it discloses data when legally required. OpenAI’s policy adds that it may retain data beyond its normal deletion window if it receives a lawful subpoena.
  • Human review. Every provider here allows some level of human review, at minimum for safety and abuse enforcement, and by default for improvement.
  • Account security. No policy prevents someone with your password from reading the journal. Session theft and shared devices are outside the document’s control.
  • Policy change. These documents get rewritten. In the month after we first published this review, Grok’s policy changed twice, on October 11 and October 27, and Google updated its Gemini privacy notice on October 13. Anthropic’s is dated September 10, 2026, and OpenAI keeps separate privacy policies by region, each with its own date. Defaults can change with notice.

None of this makes cloud assistants reckless. It makes them hosted. The right question is not “is the policy good?” It is “do I want a server copy of this at all?”

What a safer journal looks like

If the risk you care about is the server copy, the fix is to not create one. A local-first journal keeps entries in the app’s storage on your phone. There is still an app and a policy, but there is no operator-side copy of your journal to retain, review, or produce on request. If you are choosing a journaling app rather than a chatbot, our comparison of private AI journaling apps shows where Day One, Rosebud, Apple Journal, Obsidian, and Merrin process your entries, and our AI journal privacy report covers 13 apps. If you’d rather keep using a chatbot, our list of private ChatGPT alternatives includes some that never send your words to a server.

Merrin is built this way: a private personal intelligence you talk to and journal with. Conversations, journal entries, memories, and transcripts are designed to stay on your device during normal use, and the memory controls let you see, edit, and delete what it keeps — what that visibility looks like in practice is worth reading before you decide what to trust. Merrin doesn’t use the content of your journal or conversations for targeted advertising, and doesn’t sell it. The engineering deep dive on the memory loop explains how recall works, and the privacy policy covers the limited network calls, such as model downloads and purchases.

It comes with two honest costs. First, if you lose the phone and have no backup, you lose the journal; a cloud service is a useful backup until it is a liability. Second, the models that run on a phone are smaller than frontier cloud models, so they are weaker on knowledge-heavy questions, and quality varies with the device. If your journaling is mostly about saying the thing once and finding it again three weeks later, that is a trade worth understanding before you make it.

Frequently asked questions

Does ChatGPT train on my journal entries?

For a consumer account, OpenAI may use your prompts and uploaded content to train its models unless you turn off Improve the model for everyone in Settings, Data controls, or select Do not train on my content in the privacy portal. Even after you opt out, feedback you submit, such as a thumbs up or thumbs down, can be used to train models.

If I delete a ChatGPT conversation, is it gone?

OpenAI says deleted content is removed from its systems within 30 days, unless it must retain it longer for legal, safety, fraud, or financial record-keeping reasons. Temporary chats are not used for training but can be retained for up to 30 days for safety purposes.

Does ChatGPT use my conversations for ads?

On the Free and Go plans, ChatGPT shows ads, and personalized ads can draw on your current chat, past chats, and memory. Under Settings, Ad controls, you can turn off personalization or stop past chats and memory from informing ads; ads based on the current chat remain. OpenAI says advertisers get aggregated performance numbers, not your chats, history, or memories. Temporary chats show no ads, and paid plans such as Plus and Pro have none.

Does Gemini use my chats for training?

With the Keep Activity setting on, which is the default for adults, Google uses your activity to provide, develop, and improve its services, including training generative AI models, with help from human reviewers. The default auto-delete period is 18 months. With Keep Activity off, new chats are retained for 72 hours and are not used to improve Google's AI models unless you submit feedback.

Does turning off training stop the assistant remembering me?

No. Training, retention, and memory are separate controls. ChatGPT's memory settings are under Settings, Personalization; Claude's are under Settings, Your Claude; Copilot's Saved memories switch is under Settings, Personalization. Deleting a Claude conversation doesn't delete the memories Claude made from it, so remove those separately.

What is the safest way to keep an AI journal?

The journal with the least exposure is one stored on your own device, because there is no operator-side copy to retain, review, or hand over. That shifts the risk to device loss, so keep your own backup. On-device models are also smaller than frontier cloud models and weaker on general knowledge questions.