What the publisher documents
A general-purpose assistant. For some data-analysis tasks, OpenAI documents that ChatGPT writes and runs Python code in a stateful Jupyter notebook environment, can use files made available to the session, and can display tables interactively. The same documentation asks you to review the generated code, outputs and assumptions before relying on a result.
Your data
Training is controlled by one setting, Improve the model for everyone. When it is off, new conversations are not used to train OpenAI models, but they stay in your history: turning it off does not delete or hide saved chats, and archiving a chat does not change whether it can be used for training.
Temporary chats are treated separately: they do not appear in history, do not create or update memories, are not used to improve models, and may be retained for up to 30 days for safety purposes. Memory and model training are documented as two separate settings.
For organisations, OpenAI states that by default it does not use content from ChatGPT Business, Enterprise, Edu or ChatGPT for Healthcare workspaces to train its models.
Plans and what you own
The documentation names individual plans Free, Go, Plus and Pro, where signed-in users can export their data and delete their account themselves, and managed workspaces Business, Enterprise, Edu and ChatGPT for Healthcare, where export and deletion go through the workspace owner.
Documented limits
- A file can upload successfully and still not be fully analysed if it is too large, too complex, image-heavy or poorly structured.
- Exact values from image-based tables, scanned files or complex visual layouts may not be extracted reliably; a spreadsheet or text file is the documented alternative.
- The Python environment used for analysis cannot make external web requests or API calls.
- It may choose an analysis method or chart type that does not match your intent on the first try.
- Tailored advice requiring a licence, such as legal or medical advice without a qualified professional, is a prohibited use under OpenAI's policies.
What we still have to verify
What our hands-on test will have to settle: whether the analysis really computes on our own files rather than guessing, how the data controls behave in practice, and the quality of the French output next to the English.
Sources
- OpenAI Help Center, Data controls in ChatGPT
- OpenAI Help Center, Data analysis with ChatGPT
- OpenAI, Politiques d'utilisation
Publisher pages read on October 1, 2026. We do not publish prices: they change often and the official page is the only reliable one.
Our test
We have not tested ChatGPT yet. This page only reports what the vendor publishes. Our screenshots and verdict will appear here after the test.
Alternatives
- Vendor information
Claude
Assistant for long documents and careful writing.
WritePrice: to verify
- Vendor information
Gemini
Google's assistant, linked to Workspace.
WritePrice: to verify
- Vendor information
Grok
Conversational assistant that can search public X posts and the live web.
WritePrice: to verify
Guides that use it
Which AI assistant for what: comparing what the publishers actually document
Track your budget with AI without handing over your bank statements
Helping your child with homework using AI, without it becoming cheating
Asking an AI about your health: what it can do, and where it stops
Writing a complaint or administrative letter with AI, and the one thing never to let it invent
Write your team's AI charter on one page, with the text to copy
Replying to customer reviews with AI, without crossing into fake reviews
Getting cited by AI assistants: what the platforms actually document
Analysing a spreadsheet with AI: what to strip out, and how to check the numbers
How to revise for an exam with an AI assistant, step by step
Protect your personal data when you use an AI assistant
How to rewrite your CV and cover letter with AI without sounding like a robot
Summarise a meeting or a long document with AI without leaking data
Writing a product page with AI: the checklist before you publish
Write better prompts with a simple six-part method
One AI email a week. Only what changes something for you.