Is ChatGPT safe for company data? What happens to your prompts
Whether ChatGPT is safe for work data depends on the plan: Free, Plus and Pro can train on chats by default; Business, Enterprise and the API do not.
ChatGPT can be safe for company data, but only on the right plan and with the right settings. On personal Free, Plus and Pro accounts, conversations can be used to train OpenAI models by default unless the user turns that off. ChatGPT Business, Enterprise and the API are not used for training by default. On every plan, the safest rule is the same: keep customer data, credentials and secrets out of the prompt, or mask them before they leave the laptop.
Does ChatGPT train on my data? The answer by plan
| Plan | Used to train models by default? | Retention (per OpenAI) | Who controls it |
|---|---|---|---|
| ChatGPT Free, Plus, Pro (personal) | Yes, unless “Improve the model for everyone” is turned off | Chats stay in history until deleted | The individual user |
| Temporary chat (any personal plan) | No | Up to 30 days for safety purposes | The individual user |
| ChatGPT Business | No | Admin-controlled; deleted chats removed within 30 days unless law requires longer | Workspace admin |
| ChatGPT Enterprise / Edu / Healthcare | No | Admin-controlled; deleted chats removed within 30 days unless legally required | Workspace admin |
| OpenAI API | No | Up to 30 days for abuse monitoring, except listed endpoints; zero data retention available for eligible use cases | The developer or company |
Sources: OpenAI enterprise privacy and Data controls in ChatGPT, read on 6 November 2026.
What happens to a prompt after you send it?
- The text you type (and any file you upload) is sent to OpenAI’s servers over TLS.
- It is processed by the model and stored as part of the conversation, unless it is a temporary chat.
- Depending on plan and settings, it may be used to improve models. Giving thumbs-up or thumbs-down feedback can make the whole conversation eligible for training, even after opting out, according to OpenAI’s FAQ.
- On business plans and the API, OpenAI says data may be run through automated safety classifiers, and authorised staff or contractors may access it for abuse investigation, support or legal reasons.
None of this is unusual for a cloud service. The risk is not that OpenAI is careless; it is that once a value leaves your machine, you no longer control where it is stored, who can be compelled to produce it, or how long it lives.
Is ChatGPT safe for confidential information?
Treat it like any external SaaS vendor:
- Personal accounts: not suitable for customer data, health data, payment data, credentials, unreleased financials or source code with secrets. Even with training turned off, the data is stored in a personal account the company cannot see or delete.
- ChatGPT Business or Enterprise: acceptable for much internal work once legal has reviewed the DPA and retention settings. Regulated data (PHI, cardholder data) still needs a specific agreement, such as a BAA, and a policy.
- API: good controls, but your developers decide what goes into each request. One log line or support ticket pasted into a prompt can carry hundreds of customer emails.
Where do most ChatGPT leaks come from?
Most leaks do not come from the official ChatGPT Enterprise rollout. They come from people using personal accounts, browser extensions and coding assistants the company never approved. See what is shadow AI.
How do you use ChatGPT safely at work?
- Pick a business plan (Business, Enterprise or the API) so training is off by default and an admin controls retention.
- Write an AI acceptable use policy that lists approved tools and data that may never be shared. Start from our free template.
- Turn off training on personal accounts if people use them at all, and prefer temporary chats for one-off questions.
- Mask sensitive data before it leaves the device. This is the one control that works on every plan and every tool.
- Get visibility. Know which AI tools your team uses and whether sensitive data reached them.
How do you mask data before it reaches ChatGPT?
Shadow is a privacy layer that runs on your machine, free for individual users. It detects names, emails, phone numbers, card numbers, IBANs, secrets and your own organisation terms in prompts to ChatGPT, Claude, Gemini and AI APIs, replaces them with realistic, reversible fakes, and restores the real values in the answer. For ChatGPT in the browser, it works through the system proxy, masks only the prompt text and restores your values in the answer; TLS is terminated only on your own machine (how proxy mode works).
On macOS or Linux (Windows and other options):
curl -fsSL https://hub.getshadow.dev/core/install.sh | shFor teams, Shadow Hub shows which AI tools your people use and what Shadow masked or blocked, per user and per tool, at the sharing level your organisation chooses, down to counts only. Hub is $3 per user per month, with a 30-day free trial and no card required. See Pricing.
Disclaimer: This guide describes risks and technical controls. It does not make your AI use compliant with any law or standard. Involve your security, privacy and legal teams.
Frequently asked questions
Does ChatGPT train on my data?
On personal Free, Plus and Pro accounts, yes by default, unless you turn off "Improve the model for everyone" in Settings, Data controls. Business, Enterprise, Edu, Healthcare and API data is not used for training by default.
Is ChatGPT Enterprise safe for confidential data?
It has SOC 2 Type 2, admin-controlled retention, SSO and no training by default, per OpenAI. Whether it is safe for your data depends on your contracts and data classes; regulated data needs a specific agreement.
Does temporary chat keep my data private?
Temporary chats are not used for training and do not appear in history, but OpenAI may keep them for up to 30 days for safety purposes.
Can my employer see my ChatGPT chats?
On ChatGPT Business, workspace admins can view, export and delete members' conversations; on Enterprise, admins can access an audit log through the Compliance API. On a personal account, your employer cannot see them, which is exactly why personal accounts are a data-leak risk.
Is it safe to paste API keys or passwords into ChatGPT?
No. Treat any secret that reached an AI tool as exposed and rotate it. Shadow detects common secret formats and masks them before they leave your machine.