AI tools and client data
Is Microsoft Copilot safe for client data?
Short answer: Whether Microsoft Copilot is safe for client data turns on which plan your staff are on, which account they signed in with, and what you have told them they may put in it — not on the product itself. On Personal Microsoft account, we did not find a statement about model training in the pages we read; on Microsoft 365 Copilot and Copilot Chat on a work account, Microsoft says it does not train its models on what staff type, by default. Every fact on this page is Microsoft's own, quoted from its own pages and dated, and where we did not find a statement this page says so rather than filling the gap. Obsidian Ridge's free 30-day AI security assessment shows which AI tools your staff actually use.
See which AI tools your staff usePlan by plan, in Microsoft’s own words
The assistant inside Word, Outlook, Excel, Teams and the Copilot Chat panel, answering from the files and mail a signed-in person can already open.
| What you are asking | Personal Microsoft account | Microsoft 365 Copilot and Copilot Chat on a work account |
|---|---|---|
| Trains the vendor's models by default | Not found in the pages we read. | No. Microsoft says prompts, responses and the Graph data behind them are not used to train the foundation models.“Prompts, responses, and data accessed through Microsoft Graph aren't used to train foundation LLMs, including those used by Microsoft Copilot.” source, read |
| How to turn training off | Not found in the pages we read. | Not found in the pages we read. |
| How long data is kept | Not found in the pages we read. | Your own retention policy governs it, through Purview, like the rest of your Microsoft 365 content.“Admins can also use Microsoft Purview to set retention policies for the data related to chat interactions with Copilot.” source, read |
| What an admin can see or control | Not found in the pages we read. | Copilot inherits your existing permissions, which is the catch: anything over-shared internally is now one question away.“Microsoft Copilot only surfaces organizational data to which individual users have at least view permissions.” source, read |
| Single sign-on | Not found in the pages we read. | Not found in the pages we read. |
| Business Associate Agreement | Not found in the pages we read. | Microsoft lists HIPAA among Copilot's compliance offerings. The page states no BAA itself; the agreement lives in your Microsoft terms.“Microsoft Copilot provides broad compliance offerings and certifications, including GDPR, ISO 27001, HIPAA, and the ISO 42001 standard…” source, read |
| Where the data is processed | Not found in the pages we read. | Outside the EU, Microsoft says a query may be processed in another region.“Customers outside the EU may have their queries processed in the US, EU, or other regions.” source, read |
An empty cell is not a reassurance and not a criticism. "Not found in the pages we read." means the Microsoft pages listed under Sources, read on the dates shown, do not answer it — so there is nothing for us to repeat, and something for you to ask Microsoft before client information goes near it.
What this means for
The duty does not change because the tool is new. What changes is where the client information ends up.
- Law firms. A client confidence pasted into a personal account has left the firm’s control. Our law firm page sets out the duties we build evidence for; the decision here is whether a personal account may touch a client matter at all.
- Dental and medical practices. Microsoft publishes a BAA position for this product, which is where that conversation starts. See dental practices.
- CPA and accounting firms. Your written information security plan has to describe the tools people actually use, not the ones you approved. See accounting firms.
- Financial advisers. Client account data in a personal assistant is the exposure; see financial advisers.
- Software teams. A coding assistant reads the repository and can run commands, so the question is what it can reach, not what it was asked. See what AI agents on work laptops can reach.
How to tell which version your staff use
The plan is the whole answer, and you cannot tell from the outside. Three ways to find out, in order of how much they prove.
1. Ask
Quick, free, and incomplete. People forget the tool they installed once and the account they signed in with.
2. Look at the billing and the admin console
This finds what the business pays for. It does not find a personal account, which is the one that matters here.
3. Look on the computers
A tool somebody installed themselves never appears in an IT list. Obsidian Ridge’s free 30-day AI security assessment takes a quiet inventory for 30 days and shows what each tool could reach, then a CISSP goes through it with you. See the assessment, or read a sample set of findings.
Settings to check today
From Microsoft’s own documentation. Each one is a change somebody can make this afternoon.
- Fix sharing before you switch Copilot on. It answers from what people can already open, including anything over-shared.“It's important that you're using the permission models available in Microsoft 365 services, such as SharePoint” source, read
- Decide whether staff may send feedback, which is a separate admin control.“Customers can manage feedback through admin controls.” source, read
- Review which agents are allowed: an agent reaches outside Microsoft 365 with its own terms.“Admins have full control to select which agents are allowed in their organization.” source, read
Worth knowing
- Copilot Chat on a work account inherits the tenant's existing retention policy rather than needing its own.“The same retention policies used for Microsoft Copilot can be used for Copilot Chat.” source, read
- A staff member can clear their own Copilot history, which is not the same as your retention policy.“Your users can delete their Copilot activity history, which includes their prompts and the responses Copilot returns” source, read
Questions we get asked
Is Microsoft Copilot safe for client data?
Whether Microsoft Copilot is safe for client data turns on which plan your staff are on, which account they signed in with, and what you have told them they may put in it — not on the product itself.
On Personal Microsoft account, we did not find a statement about model training in the pages we read; on Microsoft 365 Copilot and Copilot Chat on a work account, Microsoft says it does not train its models on what staff type, by default.
Every fact on this page is Microsoft's own, quoted from its own pages and dated, and where we did not find a statement this page says so rather than filling the gap.
Does Microsoft train its models on what our staff type into Microsoft Copilot?
On Personal Microsoft account, we did not find a statement about model training in the pages we read; on Microsoft 365 Copilot and Copilot Chat on a work account, Microsoft says it does not train its models on what staff type, by default.
On Personal Microsoft account: Not found in the pages we read. On Microsoft 365 Copilot and Copilot Chat on a work account: No. Microsoft says prompts, responses and the Graph data behind them are not used to train the foundation models.
Microsoft's own words: "Prompts, responses, and data accessed through Microsoft Graph aren't used to train foundation LLMs, including those used by Microsoft Copilot." (read 2026-11-02).
How long does Microsoft keep what we put into Microsoft Copilot?
On Personal Microsoft account: Not found in the pages we read. On Microsoft 365 Copilot and Copilot Chat on a work account: Your own retention policy governs it, through Purview, like the rest of your Microsoft 365 content.
Microsoft's own words: "Admins can also use Microsoft Purview to set retention policies for the data related to chat interactions with Copilot." (read 2026-11-02).
Can an administrator read what our staff typed into Microsoft Copilot?
Copilot inherits your existing permissions, which is the catch: anything over-shared internally is now one question away. Microsoft's own words: "Microsoft Copilot only surfaces organizational data to which individual users have at least view permissions." (read 2026-11-02).
Will Microsoft sign a Business Associate Agreement for Microsoft Copilot?
Microsoft lists HIPAA among Copilot's compliance offerings. The page states no BAA itself; the agreement lives in your Microsoft terms.
Microsoft's own words: "Microsoft Copilot provides broad compliance offerings and certifications, including GDPR, ISO 27001, HIPAA, and the ISO 42001 standard…" (read 2026-11-02).
A signed BAA is between your practice and Microsoft; we do not sign it for you, and we do not read a compliance badge as one.
How do we find out which version of Microsoft Copilot our staff are using?
Look on the computers. A tool somebody installed themselves, or signed into with a personal account, does not appear in any IT list. Obsidian Ridge's free 30-day AI security assessment takes that inventory for 30 days and shows what each tool could reach.
Sources
Microsoft’s own pages, with the date each was read. Nothing on this page comes from anywhere else.
- learn.microsoft.com/en-us/copilot/microsoft-365/microsoft-365-copilot-privacy, read
- learn.microsoft.com/en-us/copilot/privacy-and-protections, read
All 15 tools, side by side, or run the AI tool risk checker over the ones you know about.
Other tools people ask about
Same question, different answers. The ones in the same category are the ones staff swap between.
Is Google Gemini safe for client data?
General assistant · Google
Is ChatGPT safe for client data?
General assistant · OpenAI
Is Claude safe for client data?
General assistant · Anthropic