Claude & Copilot Privacy Controls Compared
Compare account type, training preferences, history, memory and connections rather than assuming Claude or Copilot is completely private.
BMR Digital · Source review 2026-11-06 · UK English
Compare equivalent account types
A fair comparison starts by identifying the specific account and product. Claude consumer accounts, organisation plans, personal Copilot and Microsoft 365 work or school experiences may have different terms and controls. A statement about one plan should not be applied to another simply because the brand is the same. Check your signed-in identity and whether an administrator manages the service. Microsoft also distinguishes updated and older Copilot experiences in its help pages, so use the guidance matching what you see. This article is a starting point for consumer privacy review, not a claim that one service is suitable for confidential business records. If your organisation supplies an approved account, follow its process before choosing a different assistant. The question is not which logo feels safer; it is what information is processed under the actual account arrangements and for which task.
Claude model-improvement preferences have their own scope
Anthropic’s consumer guidance describes a Model Improvement preference and how conversations may be used under the applicable settings and terms. Review the current privacy explanation for your account and consider whether you want eligible conversations used for model improvement. Turning a preference off does not mean that the service no longer processes your messages to answer them, nor does it remove all safety-related handling. The official guidance discusses exceptions, including trust-and-safety uses. Avoid making a blanket “Claude never uses my conversations” claim without checking the current terms and your setting. If a task contains sensitive information, minimise it before submission rather than treating the preference as a substitute for judgement. Keep workplace contracts, passwords and unnecessary customer details out unless there is an appropriately authorised process for that precise use.
Incognito and history are not identical controls
Anthropic states that incognito chats are not used to improve Claude, including when model improvement is enabled. Read the current explanation for the mode’s availability and other handling. Do not equate a conversation not appearing in ordinary history with a guarantee of no retention anywhere. The visible history, model-improvement choice and safety processes are separate subjects. If you delete a chat, consider whether any related file or output exists elsewhere. If an assistant has a memory or personalisation feature, review it separately rather than assuming chat deletion also removes remembered information. Use a short generic task when exploring a new mode so that you can understand the workflow without uploading a sensitive test document. The aim is to make an informed choice about convenience and data use, not to chase a label that sounds completely private.
Copilot has several distinct privacy settings
Microsoft’s individual Copilot privacy guidance describes controls that can include memory, shared experiences, conversation history, web search, personalised advertising and importing Edge browser data in the updated experience. Availability depends on the product and account. Review each feature you actually use, especially where it introduces information from another service. A history preference should not be assumed to control personalisation or browser-data access. Likewise, disabling one connected experience does not establish that all data connections are disabled. Use the official page corresponding to your app version and location. Work or school Microsoft 365 arrangements need their own review, including administrator policies. A familiar Microsoft account can make access convenient, but convenience is not permission to upload another person’s private records or the business’s confidential material into an unapproved personal service.
Compare connections by what they can access
List the services and files each assistant can reach, rather than comparing only the number of privacy toggles. A tool connected to mail or a drive may handle a very different amount of information from a standalone chat. Check the account associated with each connection, its permission scope and whether you still need it. Remove one-off connections after the task when appropriate, and check permissions in the original service too. For confidential workplace tasks, get approval before connecting shared document sources. For personal tasks, consider whether a clean excerpt could solve the problem without granting broad access. Do not assume that disconnecting removes every copy of previously transferred information. Provider terms and connected-service handling may both matter. This permission-based comparison is often more practical than trying to rank products on a single vague idea of privacy.
Use the same safe task to assess both services
Try a fictional appointment reminder or a general packing list in each assistant. Observe the available controls and how the answer is produced without exposing personal information. Do not compare privacy by uploading an actual bank statement to see which tool warns you first. Both services can produce incorrect answers, so check facts against appropriate sources. Keep medical, legal and financial decisions with qualified professionals where needed. If you use generated drafts for a business, insert real customer information later through the normal approved system. This gives you a consistent workflow whichever assistant you choose. A privacy comparison should include your own information-handling habits as well as the provider’s settings. Restrictive preferences can reduce certain uses of data, but they cannot fix an unnecessary upload, an unsafe connection or a decision based on an unchecked answer.
Choose a review routine, not a permanent winner
Record account type, current training preference, history choices, memory and connections for the service you use. Revisit the record when your plan, product version or work requirements change. Keep the official Anthropic and Microsoft pages linked here available for updated wording. Do not declare either service completely safe or completely private; no simple comparison captures every feature, contractual arrangement and possible task. The Personal Data in AI guide is designed to support a cross-provider audit, while the workplace guide adds staff and supplier questions. The free checklist provides a concise reminder for everyday users. BMR Digital publishes independent educational material and is not endorsed by Anthropic or Microsoft. The most useful outcome is a clear understanding of your chosen account and a repeatable habit of sharing only what a task genuinely needs.
Official sources and further checks
Provider settings can change and differ by account, plan and location. Check the current official instructions for the service you actually use. This is independent educational guidance, not a safety or legal compliance guarantee.
A printable next step
Compare Protect Your Personal Data in AI for the stated printable guidance, or start with the free checklist. Only genuine uploaded files are offered for download.
View Protect Your Personal Data in AIUseful AI safety reading
All resourcesChatGPT Privacy Settings: What Each Control Does
Training, history, memory and connected apps are separate ChatGPT controls. Review each one before sharing personal or customer information.
AI Safety for Children: A Parent's Starting Guide
Begin with age rules, private information, supervised chatbot use and calm family conversations about deepfakes and misleading answers.
AI Policy for Small Business: Simple UK Starter Guide
Set approved tools, prohibited data, human checks and incident reporting before staff use generative AI for customer work.
AI Voice-Cloning Scams: How to Check a Fake Call
Do not rely on a familiar voice alone. End the call and independently contact the person or business through a known route.
What Not to Share With AI: Practical Examples
Keep passwords, identity documents and unnecessary personal or confidential details out of AI prompts. Use fictional examples and minimise data.
Gemini Activity & Privacy Settings Explained
Review Gemini Keep Activity, deletion, connected services and account type separately; switching off activity does not mean no processing or retention.
Claude & Copilot Privacy Controls Compared
Compare account type, training preferences, history, memory and connections rather than assuming Claude or Copilot is completely private.