---
title: "Claim Ledger: Who Can Read Your AI Chats?"
description: "A thumbs-up can override your AI privacy settings. Who can see your chats, how long they're kept, and a 10-minute check."
author: "Harry Floyd"
publication: "The Durability Curve"
canonical: "https://durabilitycurve.com/claims/who-can-read-your-ai-chats/"
essay: "https://durabilitycurve.com/blog/who-can-read-your-ai-chats/"
published: "2026-11-13"
ledger_date: "2026-11-13"
claims: 114
removed_before_publication: 0
---

# Claim Ledger: Who Can Read Your AI Chats?

*A thumbs-up can override your AI privacy settings. Who can see your chats, how long they're kept, and a 10-minute check.*

Essay: https://durabilitycurve.com/blog/who-can-read-your-ai-chats/  
Ledger (canonical, cite this): https://durabilitycurve.com/claims/who-can-read-your-ai-chats/  
Published 2026-11-13 · last verified 2026-11-13

## What the essay claims

Before: "I turned off training, so my chats are private." Or: "I deleted it, so it's gone." Or: "Nobody actually reads these."

After: each app has several switches, and each covers one thing. On every app that says, some chats can be read by people: staff and contractors at OpenAI for four stated reasons, Anthropic's Trust & Safety team when a chat is flagged, Google's human reviewers for a subset of Gemini chats even with Keep Activity off. Some chats are kept for years: flagged Claude chats up to 2 years, Gemini chats a reviewer has read up to 3 years and not deleted when you delete your activity. Rating a reply hands the whole conversation over on three of the four apps: ChatGPT may train on it even if you opted out, Claude stores it for up to 5 years, Gemini's reviewed feedback and its conversation are kept up to 3 years. Copilot's updated app says it does not use feedback to train its foundation models. On ChatGPT and Gemini a shared link is readable by anyone who has it (on Claude, a public link is), and on ChatGPT and Gemini deleting it does not remove copies people saved. Memory can outlive the chat it came from. On a work account, on all four apps the employer's administrators or account owners can get at your chats in some form.

## The claim ladder

What the ladder does not reach: no row measures how often people actually read chats, how many chats are flagged or reviewed, or whether any company follows its own page. The piece reports what each company says it does. It does not test the companies.

| # | Claim | Whose behaviour it measures | Status / scope |
|---|---|---|---|
| 1 | On personal ChatGPT plans, rating a reply lets OpenAI use the whole conversation to train its models even if you opted out; on Business, Enterprise and Edu it does not | OpenAI's help pages | Vendor documentation, read 12 Nov 2026 |
| 2 | Claude stores the whole conversation for up to 5 years when you rate a reply, de-linked from your user ID | Anthropic's privacy pages | Vendor documentation; the page does not say whether the training switch changes this, and the piece does not claim it |
| 3 | Gemini keeps reviewed feedback and its conversation for up to 3 years; with Keep Activity off, only chats with no feedback stay out of training | Google's Gemini Privacy Hub | Vendor documentation |
| 4 | Copilot says it does not use feedback to train its foundation models | Microsoft's Copilot activity-history page | Vendor documentation, updated app (18 Aug 2026) |
| 5 | OpenAI personnel and service providers may access content for four stated reasons, and all access is logged | OpenAI's consumer data page | Vendor documentation |
| 6 | Anthropic employees cannot access conversations by default, except when you give feedback or a chat needs review under the Usage Policy (Trust & Safety, need-to-know) | Anthropic's privacy pages | Vendor documentation |
| 7 | Google's human reviewers read a subset of Gemini chats, including with Keep Activity off or in temporary chats; chats are disconnected from your account first | Google's Gemini Privacy Hub | Vendor documentation |
| 8 | Microsoft's updated Copilot app pages do not say whether people review chats; the older app's FAQ said there was no opt-out from human review | Microsoft's pages | Vendor documentation; older-app lines labelled as such |
| 9 | Retention: ChatGPT deleted chats within 30 days with exceptions; temporary/incognito chats up to 30 days (ChatGPT, Claude); Gemini 72 hours with Keep Activity off; flagged Claude chats 2 years (safety scores up to 7); reviewed Gemini chats 3 years, not deleted when you delete activity | The four vendors' pages | Vendor documentation |
| 10 | A shared link is readable by anyone who has it (ChatGPT, Gemini; on Claude, a public link); deleting it does not remove copies people saved (ChatGPT) | Vendor pages | Vendor documentation |
| 11 | Deleting a chat does not necessarily delete memory made from it (ChatGPT; Claude's main memory section); turning Copilot memory off does not delete saved memories | Vendor pages | Vendor documentation; Claude's page contradicts itself in a legacy section, so the advice covers both readings |
| 12 | On work accounts, admins can view or search chats: ChatGPT: OpenAI's enterprise privacy page says workspace admins can view, export and delete, and its Business help page says owners can export a member's data; Google Vault can retain, search and export Gemini chats; Claude Enterprise's Compliance API pulls chat data; Microsoft admins use Content search or Purview | Vendor pages | Vendor documentation |

## What would make it wrong

The piece is wrong if a reader who turned off training and never rated a reply still had chats used or kept in ways the piece does not name, or if rating a reply turns out to change nothing on ChatGPT, Claude or Gemini. Either would show up as a vendor page change or an independent report.

## The evidence, row by row

Labels: VERIFIED = checked against the primary source itself by a checker who did not draft the piece, usually with the exact words, where they sit and the date read · EXECUTED = a run or observation done for the piece (code, a query, a count, or a check made live in an app); the claim is what it returned; run records are not published · CHECKED = checked against its source by a separate checker in the older ledger format, which recorded no quote, locator or date · REPORTED = not verified word for word against a primary source (secondary, not openable in full, or supported only in part); the essay words it accordingly · STRUCK = drafted, checked and removed before publication · EXCLUDED = considered and deliberately left out.

### 1. On personal ChatGPT plans, rating a reply lets OpenAI use the whole conversation to train its models, even if you opted out of training

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-1
- Quote: "For individual plans (Free, Plus, and Pro), if you submit feedback on a response—for example, by selecting thumbs up or thumbs down—we may use the entire conversation associated with that feedback to improve our models, even if you have opted out of model training."
- Source: OpenAI Help, How your data is used to improve model performance (browser read; page "Updated: 23 days ago"), FAQ, "What if I submit feedback after opting out?"
- URL: https://help.openai.com/en/articles/5722486-how-your-data-is-used-to-improve-model-performance

### 2. OpenAI says the same on its data-controls page

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-2
- Quote: "If you choose to provide feedback, such as selecting thumbs up or thumbs down on a response, the entire conversation associated with that feedback may be used to train OpenAI models."
- Source: OpenAI Help, Data controls in ChatGPT (browser read; "Updated: 2 days ago"), FAQ, "What happens if I submit feedback after opting out of training?"
- URL: https://help.openai.com/en/articles/7730893-data-controls-in-chatgpt

### 3. On ChatGPT Business, Enterprise and Edu, rating a reply does not permit training on the conversation

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-3
- Quote: "Specifically, response feedback (submitting thumbs up or thumbs down) does not permit training on the associated conversation."
- Source: OpenAI Help, How your data is used to improve model performance (browser read), FAQ, "What if I submit feedback after opting out?", para 2
- URL: https://help.openai.com/en/articles/5722486-how-your-data-is-used-to-improve-model-performance

### 4. On personal ChatGPT plans, training is on by default

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-4
- Quote: "On personal workspaces with ChatGPT Free, Plus, or Pro, data sharing is enabled by default."
- Source: OpenAI Help, Data controls in ChatGPT (browser read), Turn off Improve the model for everyone, para 1
- URL: https://help.openai.com/en/articles/7730893-data-controls-in-chatgpt

### 5. Where to turn ChatGPT training off (web)

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-5
- Quote: "Select Improve the model for everyone, turn it off, and select Done."
- Source: OpenAI Help, Data controls in ChatGPT (browser read), Web (signed in), step 4 (after "Open your account menu." / "Select Settings." / "Select Data controls.")
- URL: https://help.openai.com/en/articles/7730893-data-controls-in-chatgpt

### 6. ChatGPT voice mode has its own training switch

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-6
- Quote: "Voice mode has a separate control for sharing data for model training."
- Source: OpenAI Help, Data controls in ChatGPT (browser read), Turn off Improve the model for everyone, after the Trusted Access paragraph
- URL: https://help.openai.com/en/articles/7730893-data-controls-in-chatgpt

### 7. OpenAI staff and service providers may read your content for four stated reasons

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-7
- Quote: "A limited number of authorized OpenAI personnel, as well as trusted service providers that are subject to confidentiality and security obligations, may access user content only as needed for these reasons: (1) investigating abuse or a security incident; (2) to provide support to you if you reach out to us with questions about your account, or to troubleshoot issues affecting the product experience; (3) to handle legal matters; or (4) to improve model performance (unless you have opted out)."
- Source: OpenAI Help, How OpenAI handles data in consumer services (browser read; "Updated: 24 days ago"), Authorized human access to content, para 1
- URL: https://help.openai.com/en/articles/7039943-how-openai-handles-data-in-consumer-services

### 8. OpenAI logs every access to your content

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-8
- Quote: "Additionally, we monitor and log all access to user content and authorized personnel must undergo security and privacy training prior to accessing any user content."
- Source: OpenAI Help, How OpenAI handles data in consumer services (browser read), Authorized human access to content, para 1, last sentence
- URL: https://help.openai.com/en/articles/7039943-how-openai-handles-data-in-consumer-services

### 9. OpenAI's own warning

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-9
- Quote: "Please do not enter sensitive information that you would not want reviewed or used."
- Source: OpenAI Help, How OpenAI handles data in consumer services (browser read), Authorized human access to content, para 2
- URL: https://help.openai.com/en/articles/7039943-how-openai-handles-data-in-consumer-services

### 10. Deleted ChatGPT chats go within 30 days, with exceptions

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-10
- Quote: "Cleared chats are deleted from our systems within 30 days, unless they have previously been de-identified and disassociated from your account or we have to keep them for security or legal reasons."
- Source: OpenAI Help, How OpenAI handles data in consumer services (browser read), Delete your chat history
- URL: https://help.openai.com/en/articles/7039943-how-openai-handles-data-in-consumer-services

### 11. ChatGPT temporary chats may be kept up to 30 days

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-11
- Quote: "OpenAI may retain a copy for up to 30 days for safety purposes."
- Source: OpenAI Help, How OpenAI handles data in consumer services (browser read), Temporary chat, para 1, last sentence
- URL: https://help.openai.com/en/articles/7039943-how-openai-handles-data-in-consumer-services

### 12. Anyone with a ChatGPT shared link from a personal account can read it

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-12
- Quote: "Anyone who has a link created from a personal account can view the shared conversation or scheduled task."
- Source: OpenAI Help, Sharing conversations and scheduled tasks in ChatGPT (browser read; "Updated: 22 days ago"), Check access and privacy, Personal accounts
- URL: https://help.openai.com/en/articles/7925741-sharing-conversations-and-scheduled-tasks-in-chatgpt

### 13. Where ChatGPT lists your shared links

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-13
- Quote: "For Shared links, select Manage."
- Source: OpenAI Help, Sharing conversations and scheduled tasks in ChatGPT (browser read), Manage and delete shared links, Find shared links, step 3 (after "Go to Settings." / "Select Data controls.")
- URL: https://help.openai.com/en/articles/7925741-sharing-conversations-and-scheduled-tasks-in-chatgpt

### 14. ChatGPT can delete every shared link at once

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-14
- Quote: "To remove every link, open More actions and select Delete all shared links."
- Source: OpenAI Help, Sharing conversations and scheduled tasks in ChatGPT (browser read), Delete a shared link
- URL: https://help.openai.com/en/articles/7925741-sharing-conversations-and-scheduled-tasks-in-chatgpt

### 15. Deleting a ChatGPT shared link does not remove copies people saved

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-15
- Quote: "It also does not remove a conversation or task another person already saved."
- Source: OpenAI Help, Sharing conversations and scheduled tasks in ChatGPT (browser read), Copies saved by recipients, para 2
- URL: https://help.openai.com/en/articles/7925741-sharing-conversations-and-scheduled-tasks-in-chatgpt

### 16. A shared link left out of search is still not private

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-16
- Quote: "Shared link pages are not intended for search-engine indexing, but this does not make a link private."
- Source: OpenAI Help, Sharing conversations and scheduled tasks in ChatGPT (browser read), FAQ, "Do shared links appear in search results?"
- URL: https://help.openai.com/en/articles/7925741-sharing-conversations-and-scheduled-tasks-in-chatgpt

### 17. Deleting a ChatGPT chat does not necessarily delete the memory made from it

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-17
- Quote: "Deleting a chat alone does not necessarily delete a separate saved memory created from that chat."
- Source: OpenAI Help, Memory in ChatGPT (browser read; "Updated: 12 days ago"), Delete remembered information
- URL: https://help.openai.com/en/articles/8590148-memory-in-chatgpt

### 18. To remove a ChatGPT memory, delete the memory and the chat

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-18
- Quote: "To remove information saved as a memory, delete the saved memory and the chat where you first shared it."
- Source: OpenAI Help, Memory in ChatGPT (browser read), Remove remembered information from its sources
- URL: https://help.openai.com/en/articles/8590148-memory-in-chatgpt

### 19. Where ChatGPT memory lives

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-19
- Quote: "Select Memory."
- Source: OpenAI Help, Memory in ChatGPT (browser read), Manage memory settings, step 3 (after "Open Settings." / "Select Personalization.")
- URL: https://help.openai.com/en/articles/8590148-memory-in-chatgpt

### 20. OpenAI's enterprise privacy page says ChatGPT Business admins can view, access, export and delete members' conversations

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-20
- Quote: "Workspace admins have control over workspaces and can view, access, export, and delete end user conversations in the workspace."
- Source: OpenAI, Enterprise privacy at OpenAI (browser read; "Updated: January 8, 2026"), ChatGPT Business FAQ, "Who can view conversations and chat history in ChatGPT Business?"
- URL: https://openai.com/enterprise-privacy/

### 21. OpenAI still holds some April to September 2025 user data for the New York Times case

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-21
- Quote: "we will securely store limited historical April–September 2025 user data."
- Source: OpenAI, How we're responding to The New York Times' data demands (browser read), Update on October 22, 2025, para 3
- URL: https://openai.com/index/response-to-nyt-data-demands/

### 22. Claude stores the whole conversation for up to 5 years when you rate a reply

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-22
- Quote: "When you provide us feedback via our thumbs up/down button, we will store the entire related conversation, including any content, custom styles or conversation preferences, in our secured back-end for up to 5 years."
- Source: Anthropic Privacy Center, Is my data used for model training? (dated March 16, 2026), Feedback
- URL: https://privacy.claude.com/en/articles/10023580-is-my-data-used-for-model-training

### 23. Anthropic may use that feedback to train its models

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-23
- Quote: "We may use your feedback to analyze the effectiveness of our Services, conduct research, study user behavior, and train our AI models as permitted under applicable laws."
- Source: Anthropic Privacy Center, Is my data used for model training?, Feedback
- URL: https://privacy.claude.com/en/articles/10023580-is-my-data-used-for-model-training

### 24. Anthropic de-links feedback from your user ID

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-24
- Quote: "We de-link your feedback from your user ID (e.g. email address) before it's used by Anthropic."
- Source: Anthropic Privacy Center, Is my data used for model training?, Feedback
- URL: https://privacy.claude.com/en/articles/10023580-is-my-data-used-for-model-training

### 25. Anthropic employees cannot read your Claude chats by default, with two exceptions

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-25
- Quote: "By default, Anthropic employees cannot access your conversations unless:"
- Source: Anthropic Privacy Center, How does Anthropic protect the personal data of Claude users? (dated March 16, 2026), Your Data Protection (followed by the two bullets in rows 26 and 27)
- URL: https://privacy.claude.com/en/articles/10458704-how-does-anthropic-protect-the-personal-data-of-claude-users

### 26. Exception one: you shared it as feedback

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-26
- Quote: "You explicitly consent to share your data with us as a part of giving us feedback"
- Source: Anthropic Privacy Center, How does Anthropic protect the personal data of Claude users?, Your Data Protection, bullet 1
- URL: https://privacy.claude.com/en/articles/10458704-how-does-anthropic-protect-the-personal-data-of-claude-users

### 27. Exception two: Trust & Safety review under the Usage Policy

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-27
- Quote: "only designated members of our Trust & Safety team may access this data on a need-to-know basis"
- Source: Anthropic Privacy Center, How does Anthropic protect the personal data of Claude users?, Your Data Protection, bullet 2 ("Review is needed to enforce our Usage Policy")
- URL: https://privacy.claude.com/en/articles/10458704-how-does-anthropic-protect-the-personal-data-of-claude-users

### 28. Flagged Claude chats are kept up to 2 years, safety scores up to 7

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-28
- Quote: "We retain inputs and outputs for up to 2 years and trust and safety classification scores for up to 7 years if your chat or session is flagged by our automated trust and safety systems as violating our Usage Policy."
- Source: Anthropic Privacy Center, How long do you store my data? (dated July 1, 2026), Usage Policy Violations
- URL: https://privacy.claude.com/en/articles/10023548-how-long-do-you-store-my-data

### 29. Flagged Claude chats can be used for safety work even with training off

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-29
- Quote: "If our safety classifiers flag your conversations, they may still be used to improve our internal trust and safety models, detect harmful content, enforce our policies, or advance our safety research."
- Source: Anthropic Privacy Center, How do I change my model improvement privacy settings? (dated August 3, 2026), Please note (below the settings steps)
- URL: https://privacy.claude.com/en/articles/12109829

### 30. With training on, Claude may keep de-identified data up to 5 years

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-30
- Quote: "If you allow us to use your chats or coding sessions to improve Claude, we may retain your data in a de-identified format for up to 5 years in our model training pipelines."
- Source: Anthropic Privacy Center, How long do you store my data?, If You Allow Us to Use Your Data to Improve Claude
- URL: https://privacy.claude.com/en/articles/10023548-how-long-do-you-store-my-data

### 31. Deleted Claude chats leave back-end storage within 30 days

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-31
- Quote: "Deleted from our back-end storage systems within 30 days"
- Source: Anthropic Privacy Center, How long do you store my data?, Controlling Your Chats
- URL: https://privacy.claude.com/en/articles/10023548-how-long-do-you-store-my-data

### 32. Where Claude's training switch is

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-32
- Quote: "Under Help Improve our AI models, click the button to toggle it off/on"
- Source: Anthropic Privacy Center, How do I change my model improvement privacy settings?, Modifying Your Model Improvement Settings on your Desktop or Browser, step 4
- URL: https://privacy.claude.com/en/articles/12109829

### 33. Claude incognito chats are not used for training

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-33
- Quote: "Incognito chats are not used for training."
- Source: Claude Help, Use incognito chats, What are incognito chats?
- URL: https://support.claude.com/en/articles/12260368

### 34. Claude incognito chats are kept 30 days by default

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-34
- Quote: "While incognito chats aren't saved to your chat history, they are retained for either 30 days (default), or longer in accordance with your organization's custom data retention setting (available for Enterprise plans)."
- Source: Claude Help, Use incognito chats, What are incognito chats?
- URL: https://support.claude.com/en/articles/12260368

### 35. On Claude Team and Enterprise, incognito chats are in the owner's exports

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-35
- Quote: "Incognito chats are included in organizational data exports available to account Owners."
- Source: Claude Help, Use incognito chats, Important information for Team and Enterprise plans
- URL: https://support.claude.com/en/articles/12260368

### 36. Anyone with a Claude share link can read the snapshot (post-publish 13 Nov: true of public links; the default share is now by email invite, rows 113 and 114)

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-36
- Quote: "Once a chat has been shared, anyone with the link can view the chat snapshot."
- Source: Claude Help, Share and unshare chats, Share chats
- URL: https://support.claude.com/en/articles/10593882-sharing-and-unsharing-chats

### 37. Where Claude lists shared chats

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-37
- Quote: "Users on Free, Pro, Max, or Ultra plans can review a log of shared chats by navigating to Settings > Privacy. Find the Privacy settings section and click “Manage” next to Shared chats:"
- Source: Claude Help, Share and unshare chats, Manage shared chats, para 1
- URL: https://support.claude.com/en/articles/10593882-sharing-and-unsharing-chats

### 38. Claude Team and Enterprise chats can be shared only inside the organisation

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-38
- Quote: "Users on Team and Enterprise plans can only share chats with other members of the same organization, not publicly."
- Source: Claude Help, Share and unshare chats, Share chats
- URL: https://support.claude.com/en/articles/10593882-sharing-and-unsharing-chats

### 39. Deleting a Claude chat does not remove memory made from it

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-39
- Quote: "When a conversation is deleted, including under a retention period, memory entries generated from it aren’t removed, but you can delete individual memories while memory is on."
- Source: Claude Help, Use Claude's chat search and memory to build on previous context ("Updated yesterday"), Data retention and privacy
- URL: https://support.claude.com/en/articles/11817273-use-claude-s-chat-search-and-memory-to-build-on-previous-context

### 40. Where to see and delete what Claude remembers

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-40
- Quote: "See exactly what Claude remembers about you in Settings > Your Claude by selecting "Manage" next to Your memory (or in Settings > Memory in the older Settings layout)."
- Source: Claude Help, Use Claude's chat search and memory to build on previous context, View and manage your memory
- URL: https://support.claude.com/en/articles/11817273-use-claude-s-chat-search-and-memory-to-build-on-previous-context

### 41. On Claude Enterprise, the Compliance API pulls chat data

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-41
- Quote: "The Compliance API lets your organization programmatically pull activity feed events, chat data, and file content across all your Claude deployments."
- Source: Claude Help, Access the Compliance API, Intro, para 1
- URL: https://support.claude.com/en/articles/13015708-access-the-compliance-api

### 42. Google's human reviewers read some Gemini data

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-42
- Quote: "Human reviewers (including trained reviewers from our service providers) review some of the data we collect for these purposes."
- Source: Google, Gemini Apps Privacy Hub (Privacy questions updated October 26, 2026), Information we collect as you use Gemini Apps
- URL: https://support.google.com/gemini/answer/13594961

### 43. Even with Keep Activity off or in temporary chats, Google uses chats with help from human reviewers

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-43
- Quote: "Even if your Keep Activity setting is off or you use temporary chats, Google still uses your chats to respond to you and help protect Google, our users, and the public, including with help from human reviewers."
- Source: Google, Gemini Apps Privacy Hub, How you can control what's shared with reviewers
- URL: https://support.google.com/gemini/answer/13594961

### 44. Gemini chats are disconnected from your account before reviewers see them

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-44
- Quote: "Chats are disconnected from your account before being sent to service providers."
- Source: Google, Gemini Apps Privacy Hub, How we protect your privacy in this process
- URL: https://support.google.com/gemini/answer/13594961

### 45. Deleting your Gemini activity does not delete chats reviewers have read; those stay up to 3 years

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-45
- Quote: "If you delete your activity, past chats already reviewed by service providers aren’t deleted because they aren’t connected to your Google Account. Instead, they’re retained for up to 3 years."
- Source: Google, Gemini Apps Privacy Hub, Keep in mind
- URL: https://support.google.com/gemini/answer/13594961

### 46. Gemini keeps reviewed feedback, with its conversation, for up to 3 years, disconnected from your account

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-46
- Quote: "Reviewed feedback, associated conversations, and related data are retained for up to 3 years, disconnected from your Google Account."
- Source: Google, Gemini Apps Privacy Hub, How you can control what's shared with reviewers
- URL: https://support.google.com/gemini/answer/13594961

### 47. Google says feedback needs human review

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-47
- Quote: "Human review is necessary to help identify, address, and report potential problems raised in feedback."
- Source: Google, Gemini Apps Privacy Hub, How you can control what's shared with reviewers
- URL: https://support.google.com/gemini/answer/13594961

### 48. With Keep Activity off and no feedback, Google does not use your future chats to improve its AI models

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-48
- Quote: "If Keep Activity is off and you don’t submit feedback, Google also does not use your future chats to improve its AI models."
- Source: Google, Gemini Apps Privacy Hub, How you can control what's shared with reviewers
- URL: https://support.google.com/gemini/answer/13594961

### 49. Google's own warning

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-49
- Quote: "Please don’t enter confidential information that you wouldn’t want a reviewer to see or Google to use to improve our services, including machine-learning technologies."
- Source: Google, Gemini Apps Privacy Hub, Information we collect as you use Gemini Apps
- URL: https://support.google.com/gemini/answer/13594961

### 50. Gemini's Keep Activity is on by default for adults

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-50
- Quote: "If you’re 18 or over, Keep Activity is on by default."
- Source: Google, Manage & delete your activity in Gemini Apps, Turn off Keep Activity
- URL: https://support.google.com/gemini/answer/13278892

### 51. Gemini keeps activity 18 months by default

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-51
- Quote: "By default, your Gemini Apps activity older than 18 months is auto-deleted."
- Source: Google, Manage & delete your activity in Gemini Apps, Change Auto-delete settings
- URL: https://support.google.com/gemini/answer/13278892

### 52. With Keep Activity off, Gemini still keeps chats up to 72 hours

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-52
- Quote: "Even when Keep Activity is off, your conversations will be saved with your account for up to 72 hours to allow Google to provide the service and process any feedback."
- Source: Google, Manage & delete your activity in Gemini Apps, Turn off Keep Activity
- URL: https://support.google.com/gemini/answer/13278892

### 53. Where to turn Gemini's Keep Activity off

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-53
- Quote: "Near the top, click On Turn off or Turn off and delete activity."
- Source: Google, Manage & delete your activity in Gemini Apps, Turn off Keep Activity (computer steps, after "Click Activity .")
- URL: https://support.google.com/gemini/answer/13278892

### 54. Anyone with a Gemini public link can read the chat

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-54
- Quote: "Anyone with the link can read the chat, even if you didn’t share the link with them directly."
- Source: Google, Share your chats from Gemini Apps, Share a chat
- URL: https://support.google.com/gemini/answer/13743730

### 55. Gemini can delete every public link at once

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-55
- Quote: "Tip: To delete all your public links, click Delete all links."
- Source: Google, Share your chats from Gemini Apps, Delete a public link (after "At the bottom, click Settings Your public links .")
- URL: https://support.google.com/gemini/answer/13743730

### 56. To make Gemini forget something, delete the chats that hold it

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-56
- Quote: "To delete something that Gemini remembered about you, delete all chats with this info from Gemini Apps Activity."
- Source: Google, Get personalization with memory of your past Gemini chats, Delete or correct something Gemini knows about you
- URL: https://support.google.com/gemini/answer/16598469

### 57. On a Google Workspace account, admins can keep and search Gemini chats even after you delete them

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-57
- Quote: "Regardless of whether a user's action or setting causes the deletion of Gemini app conversations, admins can retain, hold, search, and export Gemini app conversations with Vault."
- Source: Google Workspace Help, Retain Gemini app messages with Vault (last updated 2026-11-10 UTC), How data is retained
- URL: https://knowledge.workspace.google.com/vault/retention/retain-gemini-app-messages-with-vault

### 58. Workspace content is not human reviewed outside the organisation without permission

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-58
- Quote: "Your content is not human reviewed or otherwise used for Generative AI model training outside your domain without permission."
- Source: Google Workspace Help, Generative AI in Google Workspace Privacy Hub (last updated October 6, 2026), Your privacy is our priority
- URL: https://knowledge.workspace.google.com/admin/gemini/generative-ai-in-google-workspace-privacy-hub

### 59. Copilot does not use feedback to train its foundation models

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-59
- Quote: "We don't use this feedback to train the foundation models used by Copilot."
- Source: Microsoft Support, Microsoft Copilot for individuals: your activity history, Providing feedback
- URL: https://support.microsoft.com/en-gb/privacy/microsoft-copilot/activity-history

### 60. Copilot may still use feedback to improve Copilot

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-60
- Quote: "We may use customer feedback, which is optional, to improve Microsoft Copilot, just like we use customer feedback to improve other Microsoft 365 services and Microsoft 365 productivity apps."
- Source: Microsoft Support, Microsoft Copilot for individuals: your activity history, Providing feedback
- URL: https://support.microsoft.com/en-gb/privacy/microsoft-copilot/activity-history

### 61. The updated Copilot app does not train foundation models on your chats

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-61
- Quote: "Prompts, responses, and your file contents when using the Microsoft Copilot app aren't used to train foundation models."
- Source: Microsoft Support, Microsoft Copilot for individuals: your activity history, Page body (updated app)
- URL: https://support.microsoft.com/en-gb/privacy/microsoft-copilot/activity-history

### 62. The updated Copilot app arrived on 18 Aug 2026

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-62
- Quote: "An updated version of the Microsoft Copilot app for web, desktop, and mobile devices is available as of August 18, 2026."
- Source: Microsoft Support, Microsoft Copilot for individuals: your activity history, Important box, top of page
- URL: https://support.microsoft.com/en-gb/privacy/microsoft-copilot/activity-history

### 63. The older Copilot app's FAQ applies only to the older app

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-63
- Quote: "The information in this article applies only to the older version of the Microsoft Copilot app."
- Source: Microsoft Support, Privacy FAQ for Microsoft Copilot, Top of page
- URL: https://support.microsoft.com/en-gb/microsoft-copilot/privacy-faq-for-microsoft-copilot

### 64. The older Copilot app gave no opt-out from human review

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-64
- Quote: "To ensure that our services are safe and secure for everyone, an opt-out of human review is not available."
- Source: Microsoft Support, Privacy FAQ for Microsoft Copilot (older app), "Can I opt-out of having my Copilot conversations human reviewed?"
- URL: https://support.microsoft.com/en-gb/microsoft-copilot/privacy-faq-for-microsoft-copilot

### 65. The older Copilot app kept conversations 18 months by default

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-65
- Quote: "By default, we store conversation activity for 18 months."
- Source: Microsoft Support, Privacy FAQ for Microsoft Copilot (older app), "How long are my Copilot conversations saved for?"
- URL: https://support.microsoft.com/en-gb/microsoft-copilot/privacy-faq-for-microsoft-copilot

### 66. Signing in to the updated Copilot app by 31 Dec 2026 preserves eligible older-app chats

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-66
- Quote: "Sign in to the updated Copilot app by December 31, 2026 to preserve your eligible chats and content."
- Source: Microsoft Support, Copilot app updates for personal, work, and school accounts, Top of page
- URL: https://support.microsoft.com/en-us/microsoft-365-copilot/learning/changes-microsoft-copilot-app

### 67. Turning Copilot memory off does not delete saved memories

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-67
- Quote: "Turning off Copilot Memory doesn’t automatically delete already saved memories."
- Source: Microsoft Support, Microsoft Copilot for individuals: your privacy controls and choices, Memory
- URL: https://support.microsoft.com/en-gb/privacy/microsoft-copilot/privacy-controls

### 68. Where to delete all Copilot memories

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-68
- Quote: "If you want to delete all memories that Copilot has saved, select the Delete all memories button."
- Source: Microsoft Support, Microsoft Copilot for individuals: your privacy controls and choices, Memory
- URL: https://support.microsoft.com/en-gb/privacy/microsoft-copilot/privacy-controls

### 69. Where to delete all Copilot chat history

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-69
- Quote: "To delete your entire Copilot chat history, see Manage your Copilot activity history in the privacy dashboard."
- Source: Microsoft Support, Microsoft Copilot for individuals: your privacy controls and choices, Chat history
- URL: https://support.microsoft.com/en-gb/privacy/microsoft-copilot/privacy-controls

### 70. At work, Microsoft admins can search Copilot chats

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-70
- Quote: "To view and manage this stored data, admins can use Content search or Microsoft Purview."
- Source: Microsoft Learn, Data, Privacy, and Security for Microsoft Copilot (updated 2026-11-04), Data stored about user interactions with Microsoft Copilot
- URL: https://learn.microsoft.com/en-us/copilot/microsoft-365/microsoft-365-copilot-privacy

### 71. Microsoft says its Copilot services for work have opted out of the human-review abuse monitoring Azure offers

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-71
- Quote: "While abuse monitoring, which includes human review of content, is available in Azure OpenAI, Microsoft Copilot services have opted out of it."
- Source: Microsoft Learn, Data, Privacy, and Security for Microsoft Copilot, How does Microsoft Copilot use your proprietary organizational data?
- URL: https://learn.microsoft.com/en-us/copilot/microsoft-365/microsoft-365-copilot-privacy

### 72. At work, what you see in the app is not what is kept

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-72
- Quote: "Messages visible in your AI apps are not an accurate reflection of whether they are retained or permanently deleted for compliance requirements."
- Source: Microsoft Learn, Learn about retention for Copilot and AI apps (updated 2026-06-25), How retention works with AI apps
- URL: https://learn.microsoft.com/en-us/purview/retention-policies-copilot

### 73. Switching ChatGPT training off does not delete or hide your chats

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-73
- Quote: "Turning off Improve the model for everyone does not delete or hide saved chats."
- Source: OpenAI Help, Data controls in ChatGPT (browser read; "Updated: 2 days ago"), What happens to saved chats, para 1
- URL: https://help.openai.com/en/articles/7730893-data-controls-in-chatgpt

### 74. ChatGPT temporary chats are not used for training while they stay temporary

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-74
- Quote: "Temporary chats stay out of your chat history, do not create or update memories, and are not used to improve OpenAI models while they remain temporary."
- Source: OpenAI Help, How OpenAI handles data in consumer services (browser read), Temporary chat, para 1
- URL: https://help.openai.com/en/articles/7039943-how-openai-handles-data-in-consumer-services

### 75. On Gemini, with Keep Activity off, submitting feedback lets Google collect and use data beyond the rating

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-75
- Quote: "If Keep Activity is off and you choose to submit feedback, Google collects and uses the following data to help maintain and improve Gemini Apps:"
- Source: Google, Gemini Apps Privacy Hub, "How long does Google retain my temporary chats and chats I have when Keep Activity is off, and what does Google do with this data?", after the 72-hour list
- URL: https://support.google.com/gemini/answer/13594961

### 76. That data includes the last 24 hours of your chats

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-76
- Quote: "context that can help us better understand your feedback, including the last 24 hours of your chats, and"
- Source: Google, Gemini Apps Privacy Hub, Same question, list item 2 of the feedback list
- URL: https://support.google.com/gemini/answer/13594961

### 77. Gemini feedback is reviewed by people

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-77
- Quote: "Reviewed by specially trained teams."
- Source: Google, Gemini Apps Privacy Hub, "How does Google use my feedback?", first list item
- URL: https://support.google.com/gemini/answer/13594961

### 78. Anthropic may keep chats longer when the law requires it, to resolve disputes, or to act on Usage Policy violations

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-78
- Quote: "In all cases, we may retain chats and coding sessions as required by law, to resolve disputes, or as necessary to combat violations of our Usage Policy."
- Source: Anthropic Privacy Center, How long do you store my data?, Usage Policy Violations
- URL: https://privacy.claude.com/en/articles/10023548-how-long-do-you-store-my-data

### 79. On Claude, switching training off stops future training runs but cannot remove chats from training already done

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-79
- Quote: "Your data will still be included in model training that has already started and in models that have already been trained, but we will stop using your previously stored chats and coding sessions in future model training runs."
- Source: Anthropic Privacy Center, How do I change my model improvement privacy settings?, Body, after the settings steps
- URL: https://privacy.claude.com/en/articles/12109829

### 80. ChatGPT's training opt-out covers new conversations

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-80
- Quote: "After you opt out, we won’t use your new conversations to improve our models."
- Source: OpenAI Help, How your data is used to improve model performance (browser read), Services for individuals, para 2
- URL: https://help.openai.com/en/articles/5722486-how-your-data-is-used-to-improve-model-performance

### 81. Deleting a Gemini public link leaves copies people continued in their own Gemini

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-81
- Quote: "It doesn’t delete the chat from other’s Gemini Apps activity if they continue your shared chat in Gemini Apps on their own."
- Source: Google, Share your chats from Gemini Apps, What happens when you delete a public link
- URL: https://support.google.com/gemini/answer/13743730

### 82. The older Copilot app's no-opt-out human review is for suspected Code of Conduct violations

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-82
- Quote: "Limited human review is required as part of the investigation process when a violation of the Code of Conduct is suspected."
- Source: Microsoft Support, Privacy FAQ for Microsoft Copilot (older app), "Can I opt-out of having my Copilot conversations human reviewed?"
- URL: https://support.microsoft.com/en-gb/microsoft-copilot/privacy-faq-for-microsoft-copilot

### 83. The updated Copilot app's privacy pages we read do not mention human review

- Status: EXECUTED (read 2026-11-13)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-83
- Quote: "Neither page contains the word 'human' (both pages searched in full, 12 and 13 Nov 2026)."
- Source: Our search of Microsoft's two Copilot privacy pages: activity history, and privacy controls
- URL: https://support.microsoft.com/en-gb/privacy/microsoft-copilot/activity-history

### 84. If you allow Claude to use your chats for training, a small number of Anthropic staff involved in model training can access them

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-84
- Quote: "Access is limited to a small number of personnel involved in model training."
- Source: Claude Help, I would like to input sensitive data into my chats with Claude. Who can view my conversations?, Privacy protections we provide
- URL: https://support.claude.com/en/articles/8325621-i-would-like-to-input-sensitive-data-into-my-chats-with-claude-who-can-view-my-conversations

### 85. With Gemini's Keep Activity off, future chats stay out of your activity and out of training unless you send feedback

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-85
- Quote: "Your future chats won’t appear in your Activity, and won’t be used to train our AI models, unless you choose to send Google feedback."
- Source: Google, Gemini Apps Privacy Hub, If the Keep Activity setting is off:
- URL: https://support.google.com/gemini/answer/13594961

### 86. Gemini's memory of past chats needs Keep Activity on

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-86
- Quote: "This feature isn’t available when this setting is off."
- Source: Google, Get personalization with memory of your past Gemini chats, What you need ("Have Keep Activity on.")
- URL: https://support.google.com/gemini/answer/16598469

### 87. The older Copilot app's FAQ says some conversations get human review for product improvement and safety

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-87
- Quote: "Some Copilot conversations are subject to both automated and human review for product improvement and digital safety purposes."
- Source: Microsoft Support, Privacy FAQ for Microsoft Copilot (older app), Are Copilot conversations human reviewed?
- URL: https://support.microsoft.com/en-gb/microsoft-copilot/privacy-faq-for-microsoft-copilot

### 88. Gemini settings don't control processing chats into anonymised data

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-88
- Quote: "Your Gemini settings don’t control processing of your chats to create anonymized data to improve Google services."
- Source: Google, Gemini Apps Privacy Hub, Information we collect as you use Gemini Apps
- URL: https://support.google.com/gemini/answer/13594961

### 89. The updated Copilot app's privacy pages we read give no retention period for chats

- Status: EXECUTED (read 2026-11-13)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-89
- Quote: "Neither page gives a period for keeping chats: no 'retention', 'retain', 'month', 'year' or 'days' on either (both pages searched in full, 12 and 13 Nov 2026)."
- Source: Our search of Microsoft's two Copilot privacy pages: activity history, and privacy controls
- URL: https://support.microsoft.com/en-gb/privacy/microsoft-copilot/activity-history

### 90. Anthropic's second exception: review to enforce its Usage Policy

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-90
- Quote: "Review is needed to enforce our Usage Policy"
- Source: Anthropic Privacy Center, How does Anthropic protect the personal data of Claude users?, Your Data Protection, bullet 2
- URL: https://privacy.claude.com/en/articles/10458704-how-does-anthropic-protect-the-personal-data-of-claude-users

### 91. On ChatGPT Business, members' chats are not automatically visible to other members

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-91
- Quote: "In a ChatGPT Business workspace, each user has their own chat and Codex history. Other members do not automatically see those chats or Codex activity."
- Source: OpenAI Help, Managing data, sharing, and privacy in ChatGPT Business (browser read; "Updated: 17 days ago"), Privacy and data handling, para 3
- URL: https://help.openai.com/en/articles/8798634-managing-data-sharing-and-privacy-in-chatgpt-business

### 92. Business usage analytics do not let admins read chats

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-92
- Quote: "No. Usage analytics and spend controls are separate from chat transcript access."
- Source: OpenAI Help, Managing data, sharing, and privacy in ChatGPT Business (browser read), FAQ, "Does usage analytics let admins read all user chats?"
- URL: https://help.openai.com/en/articles/8798634-managing-data-sharing-and-privacy-in-chatgpt-business

### 93. Business workspace owners can export a member's data where the option exists

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-93
- Quote: "Workspace owners can export a member’s data when the Export data option is available:"
- Source: OpenAI Help, Managing data, sharing, and privacy in ChatGPT Business (browser read), FAQ, "Can I export my data from a Business workspace?"
- URL: https://help.openai.com/en/articles/8798634-managing-data-sharing-and-privacy-in-chatgpt-business

### 94. ChatGPT's memory summary is separate from saved memories, and deleting it does not delete past chats

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-94
- Quote: "Delete and turn off memory deletes the remembered information shown in the summary and turns Memory off. It does not delete past chats."
- Source: OpenAI Help, Memory in ChatGPT (browser read; "Updated: 12 days ago"), Memory summary
- URL: https://help.openai.com/en/articles/8590148-memory-in-chatgpt

### 95. To remove something from ChatGPT's memory, it has to go from every source, including the memory summary, files and connected apps

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-95
- Quote: "Also remove the information from any other sources where it appears:"
- Source: OpenAI Help, Memory in ChatGPT (browser read), Remove remembered information from its sources (list: the memory summary, saved memories, regular and archived chats, saved files available through Library or Space, connected apps)
- URL: https://help.openai.com/en/articles/8590148-memory-in-chatgpt

### 96. Anthropic keeps feedback submissions for 5 years

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-96
- Quote: "Where you have provided feedback to us (e.g. by submitting feedback through our thumbs up/down button or sent bug reports), we retain data associated with that submission for 5 years."
- Source: Anthropic Privacy Center, How long do you store my data?, Feedback Data
- URL: https://privacy.claude.com/en/articles/10023548-how-long-do-you-store-my-data

### 97. On Claude, a deleted chat is not used to train future models

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-97
- Quote: "Additionally, if you delete a chat from your chat history, it will not be used to train future models."
- Source: Anthropic Privacy Center, How long do you store my data?, If You Allow Us to Use Your Data to Improve Claude
- URL: https://privacy.claude.com/en/articles/10023548-how-long-do-you-store-my-data

### 98. Anthropic's feedback section does not say whether deleting the chat shortens the 5 years

- Status: EXECUTED (read 2026-11-13)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-98
- Quote: "The page never mentions deleting, removing or erasing anything (searched in full, 12 and 13 Nov 2026)."
- Source: Our search of Anthropic's page 'Is my data used for model training?', which holds the feedback section
- URL: https://privacy.claude.com/en/articles/10023580-is-my-data-used-for-model-training

### 99. OpenAI's model-improvement page does not say whether a chat already used for training can be taken back out

- Status: EXECUTED (read 2026-11-13)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-99
- Quote: "The page never mentions deleting, removing or withdrawing anything (searched in full, 13 Nov 2026)."
- Source: Our search of OpenAI's page 'How your data is used to improve model performance'
- URL: https://help.openai.com/en/articles/5722486-how-your-data-is-used-to-improve-model-performance

### 100. On Gemini, the rating buttons under each answer are how you give feedback

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-100
- Quote: "Below the response from Gemini Apps, click Good response or Bad response"
- Source: Google, Gemini Apps Help: Give feedback on Gemini Apps responses, Give feedback on a response, step 1
- URL: https://support.google.com/gemini/answer/13275746

### 101. When you give Gemini feedback, the associated conversation is added to it

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-101
- Quote: "When you submit feedback, the associated conversation, like your prompts and the Gemini app’s responses, is added as part of your feedback."
- Source: Google, Gemini Apps Help: Give feedback on Gemini Apps responses, Important box, top of page
- URL: https://support.google.com/gemini/answer/13275746

### 102. ChatGPT uses voice recordings (audio or video clips) for training only if you chose to share them

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-102
- Quote: "We do not use the associated audio or video clips for training unless you choose to share them for model improvement as described above."
- Source: OpenAI Help, ChatGPT Voice (browser read; "Updated: 1 hour ago"), Data controls, Model training and audio or video clips, para 2
- URL: https://help.openai.com/en/articles/20001274-chatgpt-voice

### 103. The audio and video sharing toggles sit under Improve the model for everyone

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-103
- Quote: "Free, Plus, and Pro users in personal workspaces can choose to share clips by opening Settings > Data Controls, turning on Improve the model for everyone, and then turning on Include your audio recordings or Include your video recordings."
- Source: OpenAI Help, ChatGPT Voice (browser read), Data controls, Model training and audio or video clips, para 3
- URL: https://help.openai.com/en/articles/20001274-chatgpt-voice

### 104. Clips already disassociated from your account can still be used after you stop sharing

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-104
- Quote: "If you stop sharing, new clips will no longer be used to train our models. Clips that were previously disassociated from your account may continue to be used."
- Source: OpenAI Help, ChatGPT Voice (browser read), Data controls, Model training and audio or video clips, para 5
- URL: https://help.openai.com/en/articles/20001274-chatgpt-voice

### 105. Gemini has a memory summary where you can view, edit and delete what it understands about you

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-105
- Quote: "You can use memory summary to view, edit, and delete what Gemini Apps understands about you."
- Source: Google, Gemini Apps Help: View & manage your memory summary in Gemini Apps, Opening paragraph; path under View your memory summary (Settings, Personal Intelligence, Memory summary)
- URL: https://support.google.com/gemini/answer/18617127

### 106. Gemini's memory summary is not available in the UK, the EEA, Switzerland or Nigeria

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-106
- Quote: "not available in the EEA, Nigeria, Switzerland, or the United Kingdom."
- Source: Google, Gemini Apps Help: View & manage your memory summary in Gemini Apps, What you need, Important box, bullet 2
- URL: https://support.google.com/gemini/answer/18617127

### 107. With Keep Activity off, only five connected apps stay available on Gemini

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-107
- Quote: "If Keep Activity is turned off, only the following apps are available:"
- Source: Google, Gemini Apps Help: Manage & delete your activity in Gemini Apps, When Keep Activity is off (list: Device assistance, Gemini Notebook, Messages, Phone, WhatsApp)
- URL: https://support.google.com/gemini/answer/13278892

### 108. Gmail and Drive are among the Google Workspace apps you can connect to Gemini

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-108
- Quote: "Google Workspace, which includes services like Gmail, Calendar, Drive, Docs, Sheets, Slides, Keep, Tasks, Chat, and Meet"
- Source: Google, Gemini Apps Help: About personalization with Connected Apps, Apps you can connect, bullet 1
- URL: https://support.google.com/gemini/answer/16836988

### 109. Deleting the chats alone can leave the information in a connected app, so disconnect the app too

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-109
- Quote: "If you only delete the chats, Gemini might still find the info in the connected app."
- Source: Google, Get personalization with memory of your past Gemini chats, Delete something Gemini knows about you from a connected app
- URL: https://support.google.com/gemini/answer/16598469

### 110. Gemini's memory summary needs Keep Activity on

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-110
- Quote: "Have Keep Activity on."
- Source: Google, Gemini Apps Help: View & manage your memory summary in Gemini Apps, What you need, bullet 3
- URL: https://support.google.com/gemini/answer/18617127

### 111. You change Gemini's memory summary by prompting Gemini, including to remove things

- Status: VERIFIED (read 2026-11-12)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-111
- Quote: "Use prompts to update your info."
- Source: Google, Gemini Apps Help: View & manage your memory summary in Gemini Apps, Edit your memory summary, step 3 (its example: "Remove my old phone number from your memory.")
- URL: https://support.google.com/gemini/answer/18617127

### 112. To make Gemini forget something from a connected app, Google says you need both steps: delete the chats and disconnect the app

- Status: VERIFIED (read 2026-11-13)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-112
- Quote: "You must complete both steps because:"
- Source: Google, Get personalization with memory of your past Gemini chats, Delete something Gemini knows about you from a connected app (both reasons follow: disconnecting alone leaves past chats; deleting chats alone leaves the app)
- URL: https://support.google.com/gemini/answer/16598469

### 113. On Claude, sharing defaults to inviting specific people by email, which keeps the chat private to them

- Status: VERIFIED (read 2026-11-13)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-113
- Quote: "You can share a chat with specific people by inviting them by email. This is the default way to share a chat and keeps it private to the people you choose."
- Source: Claude Help, Share a chat with specific people, Para 1
- URL: https://support.claude.com/en/articles/16762496-share-a-chat-with-specific-people

### 114. On Claude, only a public link lets anyone with the link view the chat

- Status: VERIFIED (read 2026-11-13)
- Row link: https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-114
- Quote: "If you want anyone with the link to be able to view a chat instead, see Public links for shared chats."
- Source: Claude Help, Share a chat with specific people, Para 2
- URL: https://support.claude.com/en/articles/16762496-share-a-chat-with-specific-people

## Cite

- A claim: name the original source (and its locator) first, then the row it was checked in, e.g. "<source>, <locator>. Checked in: Claim Ledger, "Who Can Read Your AI Chats?", The Durability Curve, row <id>, https://durabilitycurve.com/claims/who-can-read-your-ai-chats/#c-<id>."
- The essay: Floyd, Harry (2026). Who Can Read Your AI Chats?. The Durability Curve. https://durabilitycurve.com/blog/who-can-read-your-ai-chats/
- This ledger: Floyd, Harry (2026). Claim Ledger: Who Can Read Your AI Chats?. The Durability Curve. https://durabilitycurve.com/claims/who-can-read-your-ai-chats/
- Downloads: https://durabilitycurve.com/claims/who-can-read-your-ai-chats.csv · https://durabilitycurve.com/claims/who-can-read-your-ai-chats.json

Licence: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Reuse it, quote it or train on it, with credit to The Durability Curve and a link. Open to AI (robots.txt: search=yes, ai-input=yes, ai-train=yes).