AI Security
29 October 2026•8 min read•By Roman Silantev

Can Claude in Chrome Remove You from Data Broker Sites? A Reality Check for Australians

One viral clip says Claude wiped a creator off hundreds of data broker sites. A public retest found him on 4 and got 2 removals, with CAPTCHAs in the way and plenty of clicking. It does work, slowly. Here's AI Lab Australia's view on when it's worth doing, how to do it safely, and why the law here makes it harder than the videos suggest.

Can Claude in Chrome Remove You from Data Broker Sites? A Reality Check for Australians

Can Claude in Chrome Remove You from Data Broker Sites? A Reality Check for Australians

One viral clip says Claude wiped a creator off hundreds of data broker sites. A public retest found him on 4 and got 2 removals, with CAPTCHAs in the way and plenty of clicking. It does work, slowly. Here's AI Lab Australia's view on when it's worth doing, how to do it safely, and why the law here makes it harder than the videos suggest.

The pitch is irresistible. Ask Claude to find every people-search site listing your name, address and phone number, then tell it to go and submit the removal requests for you. Make a cup of tea. Come back invisible.

On 29 October (Sydney time) a creator posted a TikTok retest of exactly that. He stitched another creator who claimed it had removed his digital footprint from "hundreds of websites", then ran the same prompts himself.

His result was more honest and more useful than the original. Here's what happened, what it means for Australians, and how we'd run it for a client.


What the retest actually showed

The retest used two prompts, which match a set that's been widely shared online:

  1. "find every data broker and people search website displaying my personal information"
  2. "now use Chrome to go to every one of those sites and submit the removal request for me"

Claude found his details on 4 sites. Most of the sites added CAPTCHAs, so Claude got stuck and he had to step in. He tapped to confirm requests along the way. At the end, 2 removals were confirmed.

His verdict was "100% fact", which is fair. It did work. But he also called it "not the smoothest process" and "very hands on", and said someone should build an open-source tool for it.

The original claim of "hundreds of websites"? We couldn't verify it. The table that creator showed on screen listed at least eight American sites marked as removed: Spokeo, Whitepages, BeenVerified, TruthFinder, Intelius, Instant Checkmate, USSearch and PeopleFinders. Not hundreds.

So: mostly fact. Real capability, smaller scale, more supervision than the hype implies.


What Claude in Chrome is, and what it costs

Claude in Chrome is Anthropic's way of letting Claude work inside your Chrome tabs. For this job, it can open a broker's opt-out form and fill it in for you.

According to Anthropic's permissions guide, it's available on paid plans only: Pro, Max, Team and Enterprise. If you're on the free plan, this isn't an option.

It also doesn't do everything. Claude in Chrome stops for CAPTCHAs, for verification steps, and when it hits sensitive data. Guides such as this data broker removal walkthrough and this removal kit describe the same pattern. In practice, that means you're there for every CAPTCHA, every "click the link we've emailed you", and every form that wants something you'd rather not hand over.

That's not a flaw. That's the design working. But it does mean "set and forget" isn't on the menu.


The trade the videos don't mention

Here's the uncomfortable bit. To get your address off a data broker's site, you give your address to an AI agent so it can type it into the data broker's form.

Anthropic is upfront about the risks. Its guide to using Claude in Chrome safely says:

  • Prompt injection is the biggest risk. A web page can contain hidden instructions aimed at the AI, not at you. Anthropic's own words: "The risk is not zero."
  • Claude screenshots the tabs it works in, and it can't filter out sensitive content. Whatever's on the page ends up in the conversation.
  • Use a separate browser profile. Don't run it in the same Chrome profile that's signed into your email, banking and work accounts.
  • Use Manually approve for sensitive work. In that mode Claude asks before it acts.

Anthropic's overview of Claude for Chrome goes further and advises against using it on financial, legal, medical or other sensitive sites.

People-search sites aren't banks. But they're full of exactly the personal details you're trying to protect, and they're run by companies whose business is collecting more. Justice Connect, in its guide to requesting deletion, warns against giving a broker more information than necessary. That advice applies doubly when an agent is filling in the form.

None of this is a reason to panic. It's a reason to run it carefully.


The Australian reality: thinner coverage, weaker rights

Both videos are American, and so are the sites in them. Most of the well-known people-search sites are US-based. For an Australian, that cuts both ways: you may not be listed on many of them, and the ones that do list you may be hard to reach.

The law here is also different. Three things worth knowing:

There's no general right to deletion. Justice Connect is clear that in Australia you generally can't force a broker to delete your information, and overseas brokers may be out of reach entirely.

Small businesses are mostly outside the Privacy Act, with a big exception. According to the OAIC, businesses with annual turnover of $3 million or less are generally not covered. But a business that trades in personal information is covered regardless of turnover. A local broker can't hide behind being small.

There is a path, and it's slow. Justice Connect's suggested sequence is: submit the opt-out and keep a record, send a follow-up with a 14-day final deadline if you're ignored, and if the broker has an Australian link, complain to the OAIC.

We covered who the Privacy Act applies to, and what changes for businesses using AI, in our guide to the Privacy Act and AI in Australia. If you run a business that collects customer data, that one's worth ten minutes.


It doesn't stay done

One more thing the videos skip. Removal isn't permanent.

The open-source optout project's README notes that listings come back as brokers re-scrape. It also points out that some brokers share a parent company, so one opt-out can cover several sites. PeopleConnect, for example, covers Intelius, TruthFinder, Instant Checkmate and US Search.

So plan to recheck periodically. Treat this like changing passwords, not like moving house.

And a related trap: Google's "Results about you" tool removes search results, not the page itself, according to the removal kit linked above. The listing can still be there. Google just stops pointing to it.


Step by step: how we'd run it

This is the level of detail the sources support. We won't pretend there's a magic prompt that does the lot.

  1. Pick the people. Start with the staff whose details are most exposed: directors and front-desk staff, whose names are already public through the business.
  2. Check your plan. Claude in Chrome needs a paid plan (Pro, Max, Team or Enterprise).
  3. Create a separate Chrome profile. Not signed into email, banking or work systems. This is Anthropic's own advice.
  4. Set Claude in Chrome to Manually approve. You see and approve each action.
  5. Ask Claude to find the listings. The retest's first prompt is a reasonable starting point. Write down every site it finds.
  6. Ask it to submit the opt-outs. Stay at the keyboard. Solve the CAPTCHAs, handle the verification emails, and stop it if a form asks for more than it needs.
  7. Keep a log. Site, date, what was submitted, and whether it was confirmed.
  8. Follow up with a 14-day deadline. If a broker ignores the opt-out, send a follow-up giving it 14 days. If a broker with an Australian link still ignores you, the next step is a complaint to the OAIC.
  9. Recheck periodically. Listings come back.

Hypothetical example: a three-director architecture firm in Surry Hills, where the directors' home addresses have turned up on people-search sites. One afternoon, one person, a separate browser profile, Manually approve on. Each director sits in for their own run, handles their own verification emails, and the log goes in a shared folder with a reminder to recheck. That's a supervised chore, not an automation project. And it's the right size for it.


Would we deploy this for a client?

As a supervised afternoon for a client's most exposed staff: yes. Directors and front-desk people, in Manually approve mode, in a separate browser profile, with someone watching. It saves time on the form-filling, and the human stays in charge of what goes where.

On autopilot: no. Not with an agent that screenshots every tab, can be steered by hidden instructions on a hostile page, and is working on sites built to collect personal data. Anthropic itself says the risk isn't zero. We agree, and we'd rather keep a person in the loop than find out what non-zero means.

We also wouldn't sell it as "removing your digital footprint". For Australians, the honest promise is: fewer listings on some sites, for a while, with a paper trail you can take to the OAIC if a local broker won't play ball.

If you're rolling out browser agents more broadly, have a look at our list of open-source AI security tools for AI agents before you give one the keys to a real profile.


FAQ

Can Claude remove me from data broker sites? It can find listings and fill in opt-out forms using Claude in Chrome. In one public retest it found the creator on 4 sites and 2 removals were confirmed. CAPTCHAs and verification steps still need you.

Is Claude in Chrome free? No. It's on paid plans only: Pro, Max, Team and Enterprise.

Is it safe to give Claude my personal details for this? There's real risk. Anthropic says prompt injection is the biggest one and that "the risk is not zero", and Claude screenshots the tabs it works in. Use a separate browser profile and Manually approve mode.

Do Australians have a right to be deleted from people-search sites? Not generally. You can opt out, send a follow-up with a 14-day deadline, and complain to the OAIC if the broker has an Australian link. Overseas brokers may be out of reach.

Does the Privacy Act apply to data brokers? A business that trades in personal information is covered regardless of turnover, according to the OAIC. Other small businesses with turnover of $3 million or less generally aren't.

Will my details come back? Often, yes. Brokers re-scrape, so recheck periodically.


The bottom line

The claim is real but small. Claude in Chrome can do the tedious part of data broker removal, the finding and the form-filling. It can't get past CAPTCHAs on its own, it can't make an overseas broker care about Australian law, and it can't stop listings coming back.

Run it as a supervised job for the people who need it most, with Anthropic's own safety advice switched on. Skip the autopilot.

If you're a business in Sydney or anywhere in Australia working out how to use AI agents without creating a new privacy problem, our AI consulting team can help you set it up properly, and our group training covers safe use of browser agents for staff. Or get in touch and email info@ailabaustralia.com. We'll tell you straight whether it's worth doing.

Article Tags

Claude in ChromeData BrokersPeople SearchPrivacyPrivacy ActOAICPrompt InjectionAI AgentsAI SecurityAustraliaSydney

Ready to Transform Your Business?

Get expert guidance on implementing ai security solutions for your Australian business