Remove your personal info from the internet with Claude or ChatGPT

Here’s how to do what I showed in my video, with Claude or with ChatGPT. You ask it to find the websites that show your personal info, write a removal request for each one, and fill in the removal forms from a browser. Then you can have it check again every week. You won’t write any code.

Steps 1 and 2 are the same in both. Steps 3 and 4 have a Claude part and a ChatGPT part. Pick the one you pay for.

What you need

• For Claude: a paid plan (Pro, Max, Ultra, Team or Enterprise), a computer with Google Chrome, and the Claude in Chrome extension. Anthropic says it’s on paid plans only, and it doesn’t work in other Chromium browsers or on phones. Scheduled tasks, for step 4, are on the same paid plans.
• For ChatGPT: a paid plan. Filling in the forms uses ChatGPT’s cloud browser, which is part of Work. OpenAI says it’s available on paid plans in supported regions, not on Free or Go. If you don’t see Work, that’s the first thing to check.
• About 20 minutes the first time, and a few minutes each week after that. That’s my estimate, not Anthropic’s or OpenAI’s.
• Your full name and the city you live in.

Step 1: find the sites that list you

1. Open claude.ai or chatgpt.com and start a new chat.
2. Make sure web search is on.
   • Claude: click + in the lower left of the chat box and click Web search, so a checkmark appears next to it. If you have the new Claude experience there’s no switch: Anthropic says Claude searches the web when it helps.
   • ChatGPT: turn on web search in the chat.
3. Paste this, with your own name and city:

Search the internet for [my full name], [my city]. List every website that shows my personal info, such as my address or phone number, and give me the link to remove it on each one.

The AI has to be told who “me” is. Without your name and city it searches for nobody.

4. Read the list. Open a few of the links yourself to check they’re real before you trust them. Claude cites its sources so you can check them, and both Anthropic and OpenAI say to check important results against the sources.

What you’ll see depends on where you live. Sites that sell addresses and phone numbers are mostly a US thing, so people elsewhere may find fewer.

If you screen-record this, blur your address and phone number first.

Step 2: have it write each request

In the same chat, paste:

Write a removal request for each of these sites, using that site's own removal link or form. Keep each one short and specific.

Read a couple. They contain your real details, so make sure they say what you want and nothing more.

Step 3: fill in the forms from a browser

This is where the two differ most.

• Claude in Chrome works in your own Chrome, with the accounts you’re already signed in to. Anthropic says so. It needs your computer, with Chrome open.
• ChatGPT’s cloud browser runs on a separate computer in the cloud. OpenAI says it doesn’t use your tabs, history, saved passwords or sign-ins. It works from the web or from your phone.

With Claude: Claude in Chrome

1. Add the extension. In Google Chrome, open Claude in the Chrome Web Store, click Add to Chrome, and sign in with your Claude account. Pin it: click the puzzle piece icon, then the pin next to Claude. Allow the permissions it asks for. (Anthropic’s setup page.)
2. Think about which Chrome profile. Claude sees what’s on the pages it opens. Anthropic recommends a separate Chrome profile that isn’t signed in to sensitive accounts.
3. Open it. Click the Claude icon in Chrome’s toolbar. A side panel opens next to the page.
4. Make it ask you first. In the drop-down on the chat box, choose Manually approve. Claude then asks before each action, so nothing gets submitted until you say yes. The side panel starts in Automatically approve, where Claude’s own safety check decides what to stop, not you. Don’t use Skip all approvals for this.
5. Paste:

Go to each of these sites, fill in its removal form with my details, and ask me before you submit each one.

Then paste in the list and the requests from steps 1 and 2.

6. Claude opens the sites in its own tabs, grouped in a different color, so you can tell them from yours. It keeps working if you switch tabs, as long as Chrome is open.

Where it stops and waits for you

• It asks for each site. You’ll see Allow this action, Always allow actions on this site or Decline. Anthropic calls allowing one action the safest choice, and says to use Always allow only for sites you completely trust.
• It asks before typing personal details. Anthropic says that in every mode, Claude asks for your explicit permission before it enters potentially sensitive information into a website. A removal form is full of it, so expect to approve a lot.
• It won’t do CAPTCHAs. Bypassing them is on Anthropic’s list of things Claude won’t do. Solve it yourself, then let it carry on.
• It won’t create accounts for you. That’s also on the list. If a site wants an account before it will remove you, do that one yourself.
• It won’t act on instructions in emails or web pages. Many sites email you a link to confirm the removal. Expect to click those yourself.
• Stop it if it starts doing something you didn’t ask for, like asking for unrelated information. Anthropic says to stop the task right away.

When it finishes, ask it for a list: which sites it sent a request to, which it couldn’t finish, and which still need something from you.

With ChatGPT: the cloud browser

1. Open ChatGPT on the web or on mobile and start a task in Work.
2. Paste:

Go to each of these sites and submit the removal request for me.

Then paste in the list and the requests from steps 1 and 2.

3. ChatGPT opens its own browser, on a separate computer in the cloud, and fills in forms there.
4. Follow along in the chat and answer what it asks. You don’t select the cloud browser yourself. ChatGPT picks it when it needs it.

Where it stops and waits for you

This is not a “set it and walk away” step, and OpenAI says so:

• It asks before a new website. By default ChatGPT asks permission before it opens a website for the first time. With a long list, that’s a lot of questions. In Settings > Cloud browser you can choose Always ask, Auto approve (it only stops if the address looks unsafe) or Always allow. OpenAI doesn’t recommend Always allow. Allowing a site doesn’t skip the check on important actions.
• It’s designed to ask before actions that are hard to reverse. Read what it’s about to send before you say yes.
• It stops for sign-ins and extra steps. It pauses when it needs your input, sign-in or confirmation. Some tasks need a final step you do yourself.
• Some sites will refuse it. OpenAI says some websites block automated browsers, even if the same site works in your own browser. For those, open the page yourself and send the request from your own browser.
• You can take over. If it gets stuck, ChatGPT asks you to take over the cloud browser, or you can ask it for the link.
• Stop it if it opens the wrong site or uses wrong information.

Never type passwords, codes or payment details into the chat. Use the secure sign-in form when it offers one.

When it finishes, ask it for a list: which sites it sent a request to, which it couldn’t reach, and which still need something from you.

Step 4: check again every week

Sites take time, some say no, and some put your info back. So have the AI look again every week and tell you what’s left. In both apps, ask for a check and a report. Then send any new requests yourself, the way you did in step 3.

Every week, search the internet again for [my full name], [my city]. Tell me which of these sites still show my info, and list any new ones with their removal link. Don't submit anything.

Paste your list of sites after it.

With Claude: a scheduled task

• From a chat: with the new Claude experience, you can make a scheduled task from any conversation. Say what to do and how often, answer Claude’s questions, check the name, schedule and instructions, and click Schedule.
• Or from the Scheduled page: click Scheduled in the left sidebar, then New task. Choose Create with Claude, or Set up manually and pick how often: hourly, daily, weekly, on weekdays, or manually.

What to know before you rely on it:

• It runs in the cloud, so it runs even when your computer is asleep. Anthropic says a task that needs your local files or apps only runs locally, and that tasks that need your browser need the Claude desktop app open.
• Anthropic says not to schedule the sending. Its safety page says not to schedule tasks that send messages on your behalf or take other actions that are hard to undo. That’s why this task only checks and reports.
• You can pause, resume, delete or run it now from the Scheduled page, and edit its instructions or how often it runs.
• It counts against your plan. Anthropic says everything you do with Claude counts toward your plan’s usage limits.
• Review what it finds after each run. Anthropic suggests that too.

With ChatGPT: a repeating task

You can set a finished task to repeat. Ask ChatGPT in the conversation, using the message above. You can also manage repeating tasks on the Scheduled page, at chatgpt.com/schedules. There you can edit, pause or delete them.

What to know before you rely on it:

• I couldn’t find a page saying a repeating task keeps resubmitting forms until everything is gone. What OpenAI documents is that a task can repeat daily, weekly or monthly, and that actions needing your approval pause it.
• It counts against your plan. For agent mode, OpenAI says each run of a scheduled task counts toward the monthly limit: 40 messages on Plus and 400 on Pro. Whether a Work task counts the same way isn’t spelled out, so check your usage after the first week.
• There’s a cap on active tasks: 5 on Plus and 15 on Pro.
• A task may pause. It can pause if it needs something from you, or if you delete its chat. Check Scheduled if it goes quiet.

What this won’t do

Requests only reach the sites the AI finds and can get into. Sites it doesn’t find, and sites that block it, stay up. A site can also say no, or put your info back later. So this is a clean-up you repeat, not a guarantee that you’re gone from the internet.

Also remove your info from ChatGPT’s own answers

ChatGPT can also show personal information about you in its replies. OpenAI has a separate request for that:

1. Go to privacy.openai.com.
2. Click Make a Privacy Request.
3. Select Remove personal data from ChatGPT responses.
4. Fill in the details and submit.

You don’t need a ChatGPT account. Give examples of where the info appears, say whether it’s a name, email, phone number or link, and say why it should come out. Have a government-issued ID, or other proof, ready, because OpenAI says it’s required. OpenAI decides each request case by case and can say no.

This only stops it appearing in ChatGPT’s answers. OpenAI says it doesn’t remove the info from other websites or search engines. That’s what steps 1 to 4 are for.

Written on 12 November 2026 from Anthropic’s help pages on web search, Claude in Chrome, its permissions and safety, and scheduled tasks, and OpenAI’s help pages on the cloud browser, ChatGPT agent, scheduled tasks and removing personal data from ChatGPT responses. I haven’t run these steps myself while writing this guide. Both apps change fast: if a menu has moved, look for the same words in the app.