Connectors 101: Pulling Drive Files Into ChatGPT

Most teachers meet ChatGPT the same way: copy a paragraph out of a document, paste it into the chat, get an answer, repeat. It works, but it's the manual-transmission version of the tool. Your teaching materials already live somewhere — Google Drive, OneDrive, a folder of unit plans you've refined for years. Connectors are how ChatGPT for Teachers reaches into that existing storage and works from the real files, instead of making you shuttle text back and forth by hand.

That's the promise, and it's a good one. But connectors are also the moment you grant a piece of software access to your files, which makes this a place to move deliberately. This walkthrough covers what connectors actually do, how a teacher sets one up, and — the part most tutorials skip — the specific access-scope questions worth asking before you click "allow."

Key Takeaways

• Connectors replace copy-paste with direct access. Instead of pasting text, you link a storage service so ChatGPT for Teachers can read your existing files where they already live.
• Access is granted through a permission handshake, not a password. Modern connectors use OAuth-style authorization, so you're granting scoped access — not handing over your login — and that scope is worth reading.
• Ask what "connected" actually means. The three questions that matter: what exactly can it see, what does it do with what it reads, and can you revoke access later?
• Revocation is your safety valve. Any connection you can grant, you can withdraw — from the connector settings and from your storage provider's third-party access page. Know where both live before you connect.
• Least privilege beats convenience. Connect the narrowest thing that gets the job done — a specific folder over your whole Drive — and disconnect what you no longer use.

What a connector actually does

Strip away the marketing and a connector is a bridge with a gate. On one side is your storage — say, Google Drive. On the other is ChatGPT for Teachers. The connector lets the model read files from your storage so it can answer questions about them, summarize them, or use them as source material, without you manually pasting their contents.

Concretely, that means you can ask things like "using my Unit 4 folder, draft a review guide that matches how I taught it," and the model works from your actual documents rather than a generic version of the topic. It's the same grounding advantage as file upload, but standing rather than one-shot: instead of uploading a file each time, the connection persists so your materials are reachable across conversations. The reporting on the launch — from CNBC and GovTech — lists connectors alongside search and file uploads as part of the core teacher toolkit, precisely because working from existing materials is where the real time savings hide.

Setting one up, step by step

The exact screens evolve, but the shape of the flow is stable and worth understanding before you're clicking through it:

1. Open connector settings. In ChatGPT for Teachers, find the connectors or integrations area in settings. Available connectors (like Google Drive) are listed there.
2. Choose the service and start authorization. Select the storage you want to link. You'll be sent to that provider's own sign-in and consent screen — this is the important one.
3. Read the consent screen. This screen tells you what you're granting: which account, what kind of access (read-only vs. read-write), and sometimes which scope of files. Don't reflexively click through it — this is the contract.
4. Confirm and return. Once you authorize, you're returned to ChatGPT for Teachers with the connector active. Your files are now reachable in conversations.
5. Test with something low-stakes. Ask the model to list or summarize a known file to confirm it's seeing what you expect — and nothing you don't.

Step three is the one to slow down on. The consent screen is where you learn whether you're granting read-only access to a single folder or broad access to everything in the account. Teachers rushing to get to the useful part click past exactly the information they'd want later.

The three access-scope questions

Before you connect anything, ask these three questions — and don't connect until you have answers you're comfortable with. They map directly to the risks that matter.

On the second question, the relevant reassurance is OpenAI's stated default that data in the ChatGPT for Teachers workspace is not used to train its models — but "stated default" is a starting point for your own review, not a substitute for it, exactly as we argued in the FERPA breakdown. The safest teacher posture is least privilege: connect the narrowest useful scope. If you can link a single "Teaching Materials" folder instead of your whole account, do that. A connector that can only see your lesson plans can never accidentally surface the tax document three folders over.

Revocation: knowing where the off switch is

A connection you can't undo isn't a feature, it's a hazard — so before you grant access, locate the off switch. There are two, and you want to know both:

1. Inside ChatGPT for Teachers: the connector settings let you disconnect a linked service, which stops the model from reaching those files.
2. Inside your storage provider: your account's third-party access controls list every app you've authorized and let you revoke any of them. For a Google account, that's the third-party access page in your account security settings.

Knowing about the second one matters because it's the authoritative kill switch — even if you lose access to the app, you can cut the connection from your own account side. Make revocation part of your hygiene: when a unit ends, a pilot wraps, or you simply stop using a connection, disconnect it. Standing access you're not using is standing risk you're not managing.

This is also the layer where district IT has a legitimate interest. If your school is running a formal deployment, connectors are exactly the kind of thing an administrator wants visibility into — which is why the IT director's checklist treats data access as a first-class review item rather than an afterthought.

Putting it to work

Once you're connected — narrowly, deliberately — the payoff is real. Instead of hunting for a file and pasting excerpts, you point the model at your materials and let it work: "draft a parent-friendly summary of the attached unit," "find the rubric in my assessment folder and adapt it for this project," "check my three lesson plans for this week and build a coherent review day." The materials you've already invested years in become the model's raw material, which is the whole point of grounding AI in your real work rather than the internet's average.

Pair that with structure and it compounds. Generating from your connected materials through a template in Lesson Plan Studio keeps outputs in your format, and a broader education agent skills bundle covers the surrounding tasks — worksheets, exit tickets, review guides — all drawing from the same source of truth. Connect narrow, revoke often, and let your own files do the heavy lifting.

Frequently Asked Questions

Do I need to be a technical person to set up a connector?

No. The flow is a guided authorization — pick the service, review the consent screen, confirm. The one habit that separates careful setup from careless is actually reading the consent screen instead of clicking through it, and that takes attention, not technical skill.

Can ChatGPT change or delete my files through a connector?

That depends on the access level you grant at authorization — read-only access lets it read but not modify. Prefer read-only whenever your task only needs the model to use your files rather than edit them, and check the consent screen to confirm which you're granting.

If I disconnect a connector, is my data gone from ChatGPT?

Disconnecting stops future access to those files. For questions about data already processed and OpenAI's retention, rely on OpenAI's actual data terms rather than an assumption — this is one of the specifics worth confirming directly, especially for a district deployment.

Should my whole school use one shared connector setup?

That's a district decision, not an individual one. A formal deployment should define which connectors are approved and how access is scoped, which is exactly the kind of governance the pilot checklist covers. As an individual teacher, connect narrowly and follow your district's policy.

What's the difference between a connector and a file upload?

A file upload is one-time — you attach a file for a single conversation. A connector is persistent — it links a storage service so your files are reachable across conversations without re-uploading. Connectors save more time but grant standing access, which is why the scope and revocation questions matter more.

Connectors are the upgrade from copy-paste to real integration: ChatGPT for Teachers working directly from the materials you've already built. The convenience is genuine, and so is the responsibility that comes with granting access. Ask the three questions, connect the narrowest scope that does the job, and know where both off switches live before you flip the connection on. Deliberate beats fast — and with connectors, deliberate is barely slower.

Part 34 of 100 in the ChatGPT for Teachers series. Previously: Substitute Plans in Ten Minutes with ChatGPT. Next: Prompt Patterns Every Teacher Should Steal. Browse more builder insights or explore AI skills for education at aiskill.market.