Is It Safe to Put Company Data into ChatGPT, Claude or Gemini?
Whether it is safe to put company data into ChatGPT, Claude or Gemini depends on three things: which plan you are signed in to, how its privacy settings are set, and what kind of data it is. A personal account with default settings is a different situation from a business workspace your employer has set up.
The short version: on all three providers, business and API plans do not use your data to train models by default, according to their own help pages. Consumer accounts work differently, and each provider has a setting that changes what happens. Below is what each provider’s official pages say, the exact settings to check, and a simple way to decide what can be pasted. Last verified: November 1, 2026.
What each provider says
OpenAI (ChatGPT)
- Personal accounts (Free, Plus, Pro): OpenAI’s data controls article says data sharing for model improvement is enabled by default. Turning off Improve the model for everyone under Settings → Data Controls means new conversations are not used to train OpenAI’s models. Turning it off does not delete your saved chats.
- ChatGPT Business, Enterprise, Edu and the API: OpenAI says that by default it does not use inputs and outputs from these products to train its models (ChatGPT Business privacy).
Anthropic (Claude)
- Consumer accounts (Free, Pro, Max): Anthropic’s Privacy Center says chats are used to improve Claude when you choose to allow it, when conversations are flagged for safety review, or when you otherwise opt in. It also says Incognito chats are not used to improve Claude, even if model improvement is enabled.
- Retention: with model improvement on, chats may be kept in de-identified form for up to 5 years in training pipelines. Deleted chats are removed from your history immediately and from back-end storage within 30 days (How long do you store my data?).
- Commercial products (Claude for Work, the API): “By default, we will not use your inputs or outputs from our commercial products” to train models, unless you report feedback or bugs or otherwise choose to allow it (commercial training article).
Google (Gemini)
- Gemini app with a personal Google account: Google’s Gemini Apps Privacy Hub says a subset of chats are reviewed by human reviewers, including trained service providers. It asks users: “Please don’t enter confidential information that you wouldn’t want a reviewer to see.” Turning off Keep Activity stops future chats from being reviewed. Temporary chats and chats with Keep Activity off are kept with your account for 72 hours; otherwise activity is auto-deleted after 18 months by default.
- Gemini in Google Workspace (work accounts): Google’s Workspace privacy hub says Workspace does not use customer data for training models without the customer’s permission, content is not human reviewed outside your domain without permission, and interactions stay within your organization.
Side by side
| Consumer default | Setting to change | Business / API default | |
|---|---|---|---|
| ChatGPT | Used for model improvement | Improve the model for everyone → off | Not used for training |
| Claude | Used if you allow model improvement; flagged chats may be used | Model improvement setting; Incognito chats | Not used for training |
| Gemini | Subset of chats human-reviewed | Keep Activity → off | Workspace: not used for training without permission |
“Not used for training” is not the same as “not stored”. Every provider keeps data for some period to run the service, and each has its own retention rules. If retention matters for your work, read the provider’s retention page for your plan.
The plan is only half the answer
A business plan removes training by default, but the data is still leaving your company. Before pasting, ask:
- Does a contract forbid it? Client agreements and NDAs often restrict sharing information with third parties.
- Is it personal data? Data protection law may require a lawful basis and a processing agreement with the provider. Check whether your plan’s terms include one before relying on it.
- Would a leak matter? Mistakes happen: wrong account, a shared chat link, a screenshot.
- Who controls the account? On a personal account, the employee controls the settings and the chat history, and can keep it after leaving. On a business workspace, the company does. That alone is a good reason to keep work data in work accounts.
A traffic-light rule for staff
| Colour | Examples | Rule |
|---|---|---|
| Green | Public website text, published prices, your own draft emails without names, made-up sample data | Fine on any approved tool |
| Amber | Internal procedures, non-public plans, customer names in emails, sales figures | Business plan only, minimum needed |
| Red | Passwords and API keys, card or bank numbers, ID numbers, health or HR records, client files under NDA | Do not paste |
Most tasks work as well with amber or red details removed. A complaint reply does not need the customer’s address, and a formula needs only column headers, as our spreadsheet formulas guide shows. For meeting transcripts, see how to summarise meeting notes.
Settings checklist
- Find out which account you are using: personal or a workspace managed by your employer.
- On a personal ChatGPT account, decide on Improve the model for everyone in Settings → Data Controls.
- On a personal Claude account, check the model improvement setting in privacy settings, and use Incognito chats for one-off sensitive drafts.
- On a personal Gemini account, check Keep Activity, or use a temporary chat.
- For a team, prefer a business plan or the API, and write down which tools are approved. Our AI acceptable use policy template has a ready-made section for this.
What a business plan costs
If the safest option is moving a team from personal accounts to a business plan or the API, compare the costs first. ChatGPT Plus vs Claude Pro vs Google AI Pro compares the individual plans and works through when the pay-as-you-go API is cheaper.
Frequently asked questions
Does ChatGPT train on what I type?
On personal Free, Plus and Pro accounts, OpenAI says data sharing for model improvement is on by default and can be turned off with the Improve the model for everyone setting. For ChatGPT Business, Enterprise, Edu and the API, OpenAI says inputs and outputs are not used for training by default.
Does Claude train on my chats?
For consumer accounts, Anthropic says chats are used for training if you allow it in the model improvement setting, or if they are flagged for safety review. For commercial products such as Claude for Work and the API, it says inputs and outputs are not used for training by default.
Do people read my Gemini chats?
In the Gemini app, Google says a subset of chats are reviewed by human reviewers and asks you not to enter confidential information you would not want a reviewer to see. Turning off Keep Activity stops future chats being reviewed.
Is a business plan enough to make confidential data safe?
It removes training by default, but it does not make every use appropriate. Contracts, data protection law and your own policy may still forbid sharing some data, such as health records or client files, with any outside service.
What should I never paste into an AI chat?
Passwords, API keys, full card or bank numbers, government ID numbers, and health or HR records about identifiable people. Removing them rarely changes the quality of the answer.
Related tutorials
Sources
- Data controls in ChatGPT — OpenAI
- Managing data, sharing, and privacy in ChatGPT Business — OpenAI
- Is my data used for model training? (consumer) — Anthropic
- Is my data used for model training? (commercial) — Anthropic
- How long do you store my data? — Anthropic
- Gemini Apps Privacy Hub — Google
- Generative AI in Google Workspace Privacy Hub — Google