Short answer: putting business data into the free or Plus version of ChatGPT is not safe for anything confidential — those tiers can use what you type to train the model by default. The longer answer is more useful: it depends on which version you use, what you paste, and whether you’ve changed the settings. Here’s the whole picture, in plain English.
What actually happens to your data
When you type into consumer ChatGPT, your conversation can be stored, reviewed by human trainers, and used to improve the model — unless you turn that off. Business and Enterprise tiers are different: they contractually exclude your data from training and add security controls. But even then, your data still leaves your network and sits on OpenAI’s servers, under OpenAI’s terms.
The risk isn’t hypothetical. In 2023, engineers at Samsung leaked confidential source code and internal meeting notes simply by pasting them into ChatGPT to get help — the data was out of their control the moment they hit enter.
The line you should never cross
Whatever tier you’re on, some things should never go into a public AI tool:
Names, contact details, account numbers — anything that identifies a person (PII).
Statements, tax documents, and anything covered by a confidentiality agreement.
The Samsung mistake — your intellectual property becomes someone else’s training data.
Health, legal, or financial information governed by HIPAA, GDPR, or similar rules.
Making ChatGPT safer (if you must use it)
If cloud AI is the right fit for your work, you can use it responsibly:
ChatGPT Business or Enterprise excludes your inputs from training by contract. The consumer plans don’t, by default.
On consumer plans, disable “improve the model for everyone” in settings — or use temporary chats.
Tell your team in plain terms what’s never allowed to be pasted in. Most leaks are accidents, not attacks.
Assume anything you type could be seen by someone else — and don’t type what you wouldn’t send.
When cloud isn’t enough: the alternative
For genuinely sensitive or regulated work, the safest answer isn’t a better setting — it’s not sending the data out at all. Private (on-premise) AI runs on hardware you own, so nothing you type leaves your building: no per-use fees, no third-party terms. Many businesses run a hybrid setup — a private model for confidential work, cloud tools for everything else. Our local vs. cloud guide walks through which fits your situation, and financial firms have their own page on the compliance angle.
Frequently asked questions
Does ChatGPT train on my data?
On the free and Plus (consumer) plans, yes — by default your conversations can be used to improve the model, though you can turn this off in settings. ChatGPT Business and Enterprise exclude your data from training by contract.
Is ChatGPT Enterprise safe for business data?
It’s much safer than the consumer version — it excludes your inputs from training and adds encryption, access controls, and audit trails. But your data still leaves your network and lives on OpenAI’s servers, so for regulated data many businesses still choose private or hybrid AI.
Can I use ChatGPT with client data?
Not the consumer version. For client information, use a business tier with a clear internal policy — or, for regulated or highly sensitive data, a private setup where the information never leaves your control. Many firms in law, healthcare, and finance avoid putting client data in any cloud AI tool.
Can competitors see what I type into ChatGPT?
Not directly — your chats aren’t shared with other users. The real risk is leakage: on training-enabled tiers your inputs can influence the model, and confidential data you paste in is no longer fully under your control. The danger is accidental exposure, not a competitor reading your conversations.
Not sure which applies to you?
The safe path depends on your data and your work — sometimes it’s a setting, sometimes it’s a private model. Want a human read on yours? Get in touch, or open GPT_Lenny (the chat bubble in the corner) and describe your setup.