Free guide · Hands-on AI training for teams, live online or in person in NYC and Philadelphia
Is it safe to put company data into ChatGPT?
Short answer: it depends on which plan your team is on and what they paste in. Most small teams are on personal accounts, which is the risky case.
What happens to your chats, by plan
- ChatGPT Free, Plus, Pro (personal)Chats can be used to train OpenAI's models unless you turn off Improve the model for everyone in Settings, then Data controls.
- ChatGPT Business, Enterprise, and the APIOpenAI says it does not train on this content by default.
- Claude for Work and the Anthropic APIAnthropic says it does not train on this content by default. Personal Claude plans have their own settings, so check them.
Sources: OpenAI Help Center, Data controls in ChatGPT; Anthropic Privacy Center. Checked Nov 3, 2026. Vendors change terms, so check again before you rely on this.
Training is not the only risk
- Staff on personal accounts you cannot see, manage, or shut off when they leave.
- Shared chat links that anyone with the link can open.
- Confident answers with wrong numbers, dates, or names.
- Patient data in a tool your office has no BAA for.
Five rules for your team
- Company accounts only. Pick the tool and plan, and pay for it.
- Strip names and numbers when the task does not need them.
- Never paste passwords, account numbers, or patient records.
- Read before you send. A person checks every output.
- Write it down. One page everyone signs.
Get the free one-page AI policy
Practical guidance, not legal or compliance advice.
Get your team on the same page.
20-minute call. Tell me your team size and one task you want off your plate.