Quick answer
ChatGPT is an AI product from OpenAI that you chat with, by typing or talking. As of October 7, 2026, OpenAI's overview page lists help with everyday questions, explaining, drafting, searching the web, images and decisions, and its help pages ask users to verify important information.
Key takeaways
- ChatGPT is one product from one company, OpenAI. This page uses OpenAI's own pages, read October 7, 2026.
- OpenAI's pages say it may use content from individual accounts to train its models, and that temporary chats are not used to improve its models. This site's suggestion is to keep customer personal data and credentials out of ChatGPT.
- To see whether ChatGPT is searching the web, OpenAI's own check is to look for the search or deep research icons in the input bar, or for cited web sources in the reply.
- OpenAI's help page says it encourages users to approach ChatGPT critically and verify important information from reliable sources.
What is ChatGPT?
ChatGPT is a product from OpenAI that you type or talk to, for everyday questions, drafting, web search, images and decisions.
Disclosure: this page was drafted with Claude, an AI assistant made by Anthropic, a company that competes with OpenAI.
This site's clarification: it is one company's product, not a name for AI in general. OpenAI's overview page, read on October 7, 2026, says you can use it to "Get help with everyday questions and ideas, explaining concepts, drafting messaging, searching the web, generating an image, comparing options, or thinking through a decision." The same page says: "Talk with ChatGPT in real time when speaking is easier than typing."
OpenAI's 2022 launch post says: "We've trained a model called ChatGPT which interacts in a conversational way." A banner on that post adds: "This post introduced ChatGPT in 2022. ChatGPT has evolved substantially since then." This page therefore uses the post only for how OpenAI first described the product.
This page describes only what OpenAI says about ChatGPT. Model and plan names change, so it names none as current: check OpenAI's pages.
How does ChatGPT work, according to OpenAI?
OpenAI describes it as a language model whose responses come from patterns in the data it was trained on. OpenAI's "Does ChatGPT tell the truth?" help page says ChatGPT "is designed to provide useful responses based on patterns in data it was trained on." This page does not say what ChatGPT runs on today. In a research post about language models in general, OpenAI says they "first learn through pretraining, a process of predicting the next word in huge amounts of text."
OpenAI's help page "Does ChatGPT tell the truth?" says: "But like any language model, it can produce incorrect or misleading outputs."
This site's analogy, not OpenAI's: think of a colleague who drafts from what they have read, and who can also look things up when a search tool is available.
Where will you meet ChatGPT?
On the web and in apps, and through Apple's integrations. The table lists only places an OpenAI page names; it is not a full list.
| Where | What OpenAI's page says | Page |
|---|---|---|
| Website | chatgpt.com, whose page title reads "ChatGPT: Chat, Work, Create & Code with AI" | chatgpt.com |
| Apps | The overview page has a "Download the app" button, which links to chatgpt.com/download | ChatGPT overview |
| Apple devices | The page names "iOS 18, iPadOS 18, and macOS Sequoia" and says nothing about later versions (as read on October 7, 2026) | Apple integrations help page |
OpenAI's overview page also describes a feature, not a place: "Connect to your internal work files and apps like Gmail and Slack to create polished, finished deliverables and automate tasks."
OpenAI's Apple-integrations page says "You can always use ChatGPT through Apple's integrations without logging into a ChatGPT account, to ask questions, get help writing messages, or analyze an image or document." Signed out, chatgpt.com says: "Log in to get answers based on saved chats, plus create images and upload files."
On plans, OpenAI's overview page says: "ChatGPT is available in various plans for personal use and for business and enterprise."
How can you tell whether ChatGPT is searching the web?
OpenAI's help page gives a check: look for the search or deep research icons in the input bar, or for cited web sources in the reply. "To check if search is enabled, check if the search or deep research icons are enabled in the input bar when typing in ChatGPT or whether sources from the web are cited to in the response."
OpenAI's "Does ChatGPT tell the truth?" page says "While ChatGPT does not search the web unless specific tools are enabled, these tools are enabled for all models by default," and that "Depending on your plan, ChatGPT may have access to tools that help it answer with more up-to-date or verifiable information". Use the check above for what your own screen does.
What should you not paste into ChatGPT?
OpenAI's current pages tie what happens to your content to your account type and settings. This site's suggestion, not OpenAI's: keep out customer names with contact details, payment details, health or legal files, passwords and login codes, and anything a contract tells you to keep confidential. Check anything before it leaves your business.
Training. For individual accounts, OpenAI's "How your data is used to improve model performance" page says: "When you use our services for individuals, such as ChatGPT and Codex, we may use your content to train our models." The data-controls page says "Open Settings and select Data controls to manage the options available to your account," and "When Improve the model for everyone is off, your new conversations won’t be used to train OpenAI models." It adds that "Available controls depend on whether you are signed in, your plan, and your workspace settings." For work, it says that by default OpenAI does not use content from ChatGPT Business, Enterprise, Edu, or ChatGPT for Healthcare workspaces to train its models.
Deleting and temporary chats. On the data-controls page, the FAQ asks "Does turning off model training delete my chats?" and answers: "No. Delete a chat or your account separately when you want to remove content." Of temporary chats, the page says that "While they remain temporary", they "may be retained for up to 30 days for safety purposes" and "are not used to improve OpenAI models".
This site's rule, built on those sentences, not OpenAI's: turning training off does not delete your chats, so delete those separately. A temporary chat is not used to improve OpenAI's models but may be kept up to 30 days for safety purposes. Keep customer details out, as the list above says; if you must include some, use a temporary chat (kept up to 30 days) or a workspace plan whose default OpenAI states ("By default, OpenAI does not use content from ChatGPT Business, Enterprise, Edu, or ChatGPT for Healthcare workspaces to train its models."), and on an individual account check "Improve the model for everyone" first. The pages read here do not state what that setting is set to on a new account.
What are five safe first tasks, and what should you check?
This is this site's suggestion, not OpenAI's: start with tasks that need no customer data, matched to uses OpenAI's overview page names.
- Drafting messaging. Write a reply to a made-up customer question, then edit it to your voice. See also AI email responder for small business.
- Explaining concepts. Ask it to explain a term from a vendor quote in plain words, then read the vendor's document.
- Comparing options. Give it two options you already describe publicly and ask for questions to ask before choosing.
- Thinking through a decision. Describe a choice in general terms and ask what you may have missed.
- Searching the web. Ask about a public topic with search on, then open every linked source before relying on it.
The check habit. Edit every draft before it goes out, and open every linked source. For the wider setup, see the small business AI automation guide.
Next step for this week. Open Settings, then Data controls, and read which options your account shows. Then run task 1 once with invented details.
How was this page made?
This page was drafted with Claude, an AI model made by Anthropic, a company that competes with OpenAI (see the disclosure under the opening answer), and checked against each vendor's published material on October 7, 2026. OpenAI is the only vendor described.
Statements about ChatGPT rest on OpenAI pages opened that day, with quotations copied literally. The pages were read rendered in a browser from outside the United States, because OpenAI's help pages refuse plain downloads. OpenAI can change them at any time. How-it-works explanations are OpenAI's, not tested here. Not covered: plans, prices, accuracy rates, model names and other assistants.
FAQ
Do I need an account to use ChatGPT?
Not for every route. OpenAI's Apple-integrations page says: "You can always use ChatGPT through Apple's integrations without logging into a ChatGPT account, to ask questions, get help writing messages, or analyze an image or document." The "Signed-out use" section of the data-controls page begins: "When using ChatGPT without signing in, open Settings and turn off Improve the model for everyone." What logging in adds, per chatgpt.com: "Log in to get answers based on saved chats, plus create images and upload files." OpenAI's pages read here do not say where signed-out use is available, so check what your screen shows.
Can I use ChatGPT for work chats?
On an individual account, OpenAI says it "may use your content to train our models" unless you turn off Improve the model for everyone; workspace plans have the stated default that "OpenAI does not use content from ChatGPT Business, Enterprise, Edu, or ChatGPT for Healthcare workspaces to train its models."
What happens to my chat if I give a thumbs up or down?
In answer to "What happens if I submit feedback after opting out of training?", OpenAI's data-controls page says: "If you choose to provide feedback, such as selecting thumbs up or thumbs down on a response, the entire conversation associated with that feedback may be used to train OpenAI models."