Pre-departure checklist · Before typing into an AI chat window

6 Checks Before Entering Personal Information into a Free AI

Don't paste names, contact details, or customer lists into the chat window as they are. Before you press Enter, just run through these six items.

AI Cluster Map · 2026.10.29 · Checked on 2026-10-29
6 items
Checklist items
Takes about a minute to scan before you type
3 types
Information to be careful with
Your info, other people's info, internal company materials
3 sources
Sources consulted
SK Shieldus · Namuwiki · LG Uplus

Observation just before entering

Changing orbit

Check the route rules

Verdict
If even one of the six is blocked, lean toward not sending that sentence. If you swap in blanks and send it, you can get the same help.

FAQ

Is it a problem to enter just a name into a free AI?It's hard to say that a name alone will immediately cause serious harm. However, when a name is combined with contact details, workplace, or address, a specific person becomes identifiable. If it isn't truly necessary, replace it with a blank like [name], and get into the habit of not stacking multiple pieces of information in one conversation.
Is what I enter into ChatGPT used for training?The Namuwiki article on ChatGPT states that OpenAI collects data from ChatGPT users to further train and fine-tune its service. However, that is the wiki's description, and how data is handled may differ depending on the account type and settings. Check the exact conditions yourself in OpenAI's official terms, its privacy policy, and your account settings screen.
What's the safe way to summarize company materials?First, check whether your company has standards for AI use. If it doesn't, it's safer not to put customer information or internal materials into an external AI. If you must use one, erase identifying parts such as names, company names, and amounts, or replace them with fake values, and show only the structure. You should also compare the summary against the original to make sure nothing is wrong.
Cover image: why you shouldn't enter personal information into a free AI, and what to do instead
Cover image · Image: made in-house
Key points card: 5 things to check before entering anything into an AI
Key points card · Image: made in-house
Highlight box summarizing the risks of entering personal information into a free AI chat window
Highlight box · Image: made in-house

A closer look

Which star does the sentence I typed fly to?

The short answer: it doesn't stay on your device. A sentence you write in an AI chat window isn't processed only on your device; it is sent to the service company's servers.

The Namuwiki article on ChatGPT states that OpenAI collects data from ChatGPT users to further train and fine-tune its service. Pressing thumbs up or down on an answer, or writing feedback, is part of that process too.

However, this is the wiki's description based on ChatGPT. You can't assume other services are exactly the same. So it's safer to use a service on the assumption that "what I enter may be used to improve the service." Terms and settings differ by service, so check the privacy policy of the service yourself before using it.

Which information is the most dangerous reef?

Just remember three types. Information that identifies you (name, contact details, address, account number), other people's information (customers, colleagues, family), and internal company materials (contracts, sales tables, plans not yet made public).

SK Shieldus published an article laying out the privacy rules that individuals and companies should know when using generative AI such as ChatGPT. The title itself asks, "Is it okay to enter personal information?" However, this time I only checked the search summary. For the specific rules, read them yourself in the SK Shieldus guide.

LG Uplus's article on adopting AI for business also points to customer information and internal materials going into external AI as a problem. In particular, other people's information needs extra care, since sending it out amounts to doing so without their permission.

Will moving to a paid orbit make the worry go away?

There's no basis for assuming that. The LG Uplus article groups Free and Go, Plus, Pro (paid) together as "plans that an individual uses alone." It explains that each plan differs in how much you can use per day and which AI models you can use. This means the differences among individual plans are mainly usage and models.

The same article agrees that paid plans let you use more and faster than free ones, but says the reason a company should use a team or enterprise plan is not performance. The point is that you also need to look at whether things stay manageable when several people use it together.

So don't assume "I'm paying, so it must be safe." Check the data handling conditions of the plan you use separately. For pricing and limits, the official pricing page is the accurate place to check.

Can I still get AI to do the work without personal information?

Yes. What an AI needs is mostly not "whose information it is" but "the structure of the sentence and the situation." If you mask names and numbers before entering them, you can get the same help.

For example, change "Write an apology text about a delivery delay to customer Hong Gildong (010-0000-0000)" like this: "Write a 3-sentence apology text about a delivery delay to send to customer A. Leave the name and contact number blank as [name] and [number]." After you get the reply, just fill in the blanks yourself on your own computer.

The same principle applies to Excel tables. Don't upload the whole table; show only the column names and two lines of fake sample data, then ask about the formula. Your real data never leaves your computer.

What if it's for work, or if you've already entered it?

Before choosing a tool, decide "what is OK to enter." LG Uplus's article on adopting AI also covers the step of setting input rules and a way to check outputs first. If employees start using tools on their own, it becomes hard to trace where customer information has gone.

One part is easy to get wrong. Work email and internal documents feel like "my work material," so they seem fine, but if customer names or contact details are mixed in, entering them means putting personal information into an outside service. Before uploading, scan for what information is included, and if your company has an approved tool, use that instead of a personal free account.

If you have already entered something, start by recalling what you entered. If it is easily misused information, such as a resident registration number or a password, the first priority is whether you can change it. Settings for deleting chat history or turning off use for training can differ by service, so look them up yourself in that service's settings and privacy policy. If you entered company materials or customer information, don't try to handle it alone. It's best to tell the responsible department.

✦ Go see this star on the 3D map
Sources: SK Shieldus 'Is it OK to enter personal information into ChatGPT?' (search summary), Namuwiki ChatGPT, LG Uplus 'ChatGPT paid subscription: how many people at our company should get it first?' · Checked 2026-10-29

This post was translated from the Korean original. Prices, plans and dates are as checked in the original. Read the Korean original

#Free AI#Personal information#ChatGPT#AI security#Generative AI#Cautions when using AI#Privacy protection#Pre-departure checklist
HomeTrendsTools