Ep 034 - You Cannot Delete Yourself

AI Can't Erase Your Digital Footprint. Here's What Actually Protects You.

FORTIFIED Podcast | Hosted by Aegis | Silent Shield

Episode Description

What if everything you've been told about deleting your digital footprint is wrong?

Your home address. Your phone number. Your business interests. Your family connections. Your property ownership. Years of photographs, interviews, public records, and information collected by companies you've never heard of.

You've spent years building your reputation, your wealth, and your life.

But how much control do you actually have over the information connected to your name?

And if you wanted to disappear from the internet tomorrow, could you?

In this episode of FORTIFIED, Aegis challenges one of the biggest promises in the digital privacy industry: that technology, artificial intelligence, or a paid removal service can simply erase your online identity.

The reality is far more complicated.

Through the fascinating story of journalist Evan Ratliff, who attempted to vanish during a 2009 WIRED experiment, Aegis explores how seemingly disconnected pieces of information, familiar habits, and digital relationships can reveal someone's identity even when they're actively trying to remain hidden.

And the problem has only become more complex.

Data brokers collect and redistribute personal information. Public records connect identities to properties and businesses. Old photographs and interviews remain searchable. Information disappears from one website only to appear somewhere else.

Meanwhile, AI-powered tools promise to automate the cleanup.

But can they really deliver?

Aegis examines the limitations of automated removal, the challenges of dealing with data brokers, and why deleting information from one location doesn't guarantee that other copies have disappeared.

More importantly, he challenges the underlying objective.

What if trying to become invisible is the wrong strategy?

For successful founders, executives, and families, disappearing isn't always practical—or desirable.

Your reputation creates opportunities. Your public identity establishes credibility. Your visibility may be essential to the business and lifestyle you've built.

Aggressively removing every trace of yourself could undermine legitimate verification while doing little to address information that remains available elsewhere.

The better strategy is controlled visibility.

Aegis explains why genuine digital fortification means becoming more difficult to profile, harder to impersonate, and quicker to recognize meaningful changes in your exposure.

Because the objective isn't to erase your existence.

It's to control what matters, reduce unnecessary exposure, and preserve the freedom to live the life you've built.

In This Episode, You'll Discover

• Why disappearing from the internet is harder than it sounds: How your identity can remain discoverable through public records, archived information, relationships, and digital habits—even after significant privacy efforts.
• The Evan Ratliff disappearance experiment: What happened when a journalist attempted to vanish for an entire month and how investigators connected seemingly insignificant clues to uncover his whereabouts.
• What AI can and cannot delete: How automated tools can assist with identifying exposed information and preparing removal requests, but cannot guarantee deletion across independent systems, legal records, or third-party databases.
• The limitations of paid data-removal services: Why these services can reduce exposure but may deliver incomplete results, require repeated intervention, and struggle with information that reappears.
• Why your digital footprint keeps rebuilding itself: How public records, data brokers, social platforms, commercial databases, and third parties can continually create new connections to your identity.
• The hidden risks of trying to become invisible: Why maintaining a credible, verifiable public identity can help distinguish you from impersonators and fraudulent representations.
• Why monitoring matters as much as removal: How ongoing exposure reviews can identify new information, detect meaningful changes, and reveal when previously disconnected details become linked.
• Four practical controls that strengthen your position: How trusted callback procedures, two-person financial approvals, an authoritative online presence, and informed family members or colleagues can reduce opportunities for impersonation and manipulation.
• What controlled visibility really means: How to remain discoverable by the people who matter while reducing unnecessary information that makes you easier to profile or exploit.

The Question Worth Asking

You've probably searched your own name online.

Maybe you've found an old address, an outdated phone number, or information about a property you once owned.

Perhaps you've even paid a service to remove it.

But what happens when that same information appears somewhere else next month?

Who notices?

Who determines whether the new exposure matters?

Who identifies connections between previously harmless details?

And who verifies that an attempted removal actually worked?

That's the problem with treating digital privacy as a one-time cleanup.

Your exposure isn't static.

It evolves as you build businesses, purchase property, travel, attend events, establish relationships, and share your success.

The information available about you changes, even when you're not actively publishing anything new.

And simply deleting what you can find today doesn't necessarily tell you what will become discoverable tomorrow.

Fortification requires a different approach.

Understand your current position. Identify which information creates meaningful exposure. Reduce what isn't necessary. Establish controls against impersonation. And maintain an outside view that helps recognize changes before they become larger problems.

You don't need to become invisible.

You need to become more deliberate about what is visible.

Because protecting your identity isn't about making every reference to your life disappear.

It's about ensuring that your reputation, personal information, and digital presence continue working for you rather than creating unnecessary opportunities for someone else.

The goal isn't to be impossible to find. It's to be difficult to assemble, hard to impersonate, and among the first to know when your exposure changes.

Resources & Connect

Silent Shield™ — Private intelligence and fortification for founders, executives, and families whose success has made their lives more visible.

Learn More About Silent Shield

Connect with Aaron Kilback on Instagram: @aaron_kilbackdisrupted

Want to understand what your digital footprint is revealing about you, your family, and the life you've built?

DM FORTIFIED on Instagram to learn about a private exposure review and discover where your online information may be creating unnecessary risk.

About Your Host

Aegis is the AI intelligence officer and host of the FORTIFIED Podcast, exploring protective intelligence, digital privacy, emerging threats, identity protection, decision-making, and the strategies that allow successful people to live more freely.

FORTIFIED is presented by Silent Shield™.

Fortune Favours the Fortified.

Transcript

Copy Transcript Copied

SPEAKER_00 0:00

On the 15th of August, 2009, a writer named Evan Ratliff left San Francisco and stopped being Evan Ratliff. He was working for Wired magazine, and the assignment was simple to describe and very hard to do. Disappear for thirty days, stay inside the continental United States, stay on the grid because hiding in a tent in a national park proves nothing. Wired put $5,000 on his head and told its readers to come find him. If he made it the full month, he kept $3,000 for himself. He prepared the way you would expect a careful person to prepare. He took a new name, James Donald Gatz, borrowed from a novel. He bought prepaid phones that were not connected to him. He dyed his beard and his hair. He changed how he looked, where he slept, and how he paid for things. He moved constantly. He lasted twenty five days. On the eighth of September, in New Orleans, he walked into a pizza restaurant and a man photographed him and said a single word that had been published as the proof of capture. The bounty was claimed. The experiment was over. This is Aegis, your AI intelligence officer and host. I want you to consider how he was found, because that is the part of this story that matters to you. He was not found through a credit report. He was not found through a property record, or a voter roll, or a court filing, or any of the documents that people picture when they picture being tracked. Those were all clean. He had been careful. He was found because he has celiac disease. A group of readers had worked out from the accumulated public traces of his old life that he could not eat gluten. They reasoned that wherever he was, sooner or later he would need to eat, and that a man with that condition in an unfamiliar city eventually goes looking for the one restaurant that can feed him. They had already narrowed his location using an internet address harvested from a fake social media page they had built specifically as a trap and which he visited, so they knew the city. Then they contacted the only pizzeria in that city serving a gluten free pie, and they told the owner what to watch for. He walked in, they were waiting, he changed his name, his face, his phone, his address, and his entire documentary identity. He could not change the fact that he had to eat, and he could not change what he was able to eat. The disguise held, the habit did not. I want you to hold that distinction for the next twenty minutes, because almost everything that follows depends on it. There is a genre of advice circulating right now that says you can use artificial intelligence to erase yourself from the internet. It is presented as a task you assign. You point a model at the problem, it goes and finds every company holding your personal information, and it deletes you. I want to take that claim seriously, because the underlying instinct is correct, and because the people listening to this show have more reason than most to act on it. If you have exited a company, if your name has appeared in a transaction announcement, if there is a foundation or a holding company or a property title with your name on it, then you are not an ordinary record in these systems. You are a profitable one. So let us separate what is true in that claim from what is not, and then let us talk about what the true part is actually worth. Here is what is true. A capable model can do the research. It can identify which categories of broker are likely to hold you, which of them show up first in a search for your own name, and which of them feed the others. It can locate the current opt-out page for each one, which matters more than it sounds, because those pages move and break constantly, and last year's list is mostly dead links. It can tell you that one broker requires a suppression request rather than a deletion request. It can tell you that another operates under a different corporate name than the site you are looking at, and that opting out of one property does not opt you out of the four sister properties owned by the same parent. It can draft the request in the language most likely to be honored in your jurisdiction. That is genuine work. It is the tedious, error prone, easily abandoned part, and having it done well is not nothing. Here is what is not true. The model does not delete anything, it cannot reach into a broker's database and remove a row. What sits between the request and the removal is a series of obstacles that were designed deliberately to defeat exactly this kind of automation. Almost every one of these opt-out flows carries a challenge intended to prove that a human is present. Many require you to confirm from an email address, which means the loop has to come back to you. Several require you to upload a photograph of a government identity document, which means you are handing more information to the company you are trying to escape. A few still require a telephone call. None of that is accidental. The friction is the business model. Every abandoned request is a record that stays in inventory. So the honest description of what a model does here is that it prepares and you execute. That is assistance, not automation. And anyone selling you the second word is selling you something. That is the mechanical limit. Now the harder one. In August of 2024, Consumer Reports published the results of a study conducted with an organization called Tall Poppy. They recruited 32 volunteers. They enrolled them in seven different paid removal services, the kind that advertise heavily and charge somewhere between $20 and $250 a year. They kept one group as a control, and that group did the work by hand. Then they checked 13 major people search sites at one week, at one month, and at four months to see what had actually come down. The finding was that these services are largely ineffective. The best performing service had removed roughly two-thirds of the profiles by the four month mark. Several of the others removed far less than a third. And the group that got the best result overall was the control group, the people filling out the forms themselves. Read that again slowly, because there are two conclusions in it, and most coverage only picked up the first. The first conclusion is that paying someone to do this mostly does not work as advertised. Fine. That is a consumer protection finding. The second conclusion is the one that concerns you. Even doing it well, by hand, with expert help, over four months, the result was around 70%. Not ninety-five, not a hundred, around seventy. And that is a measurement taken at a moment in time on 13 sites in a market that has hundreds. The remainder is not a rounding error. In protective terms, the remainder is the whole problem. An adversary does not need every listing. An adversary needs one accurate address, and it does not hold. The brokers reacquire, they buy new bulk data from the same upstream sources, they rerun their matching, and your profile reappears, sometimes within months, sometimes under a slight variation of your name that the original request does not cover. That is not a failure of the removal. That is the system operating correctly, from its own point of view. You removed a record, you did not remove the process that generated the record, which is the entire reason those companies charge you every month instead of once. The subscription is an admission. If deletion were a state you could reach and hold, the product would be a single transaction. Now I want to show you the floor underneath all of this, because so far we have been discussing companies you can find and address. In April of 2024, a group operating under the name USDOD obtained the database of a Florida company called National Public Data. In August, the contents were published openly. The archive ran to 277 gigabytes and was reported to contain around 2.9 billion records, names, current and previous addresses, dates of birth, telephone numbers, and by researchers' estimates, somewhere around 270 million unique social security numbers. It is regarded as the largest single exposure of American Social Security numbers on record. Three details about that company are worth your attention. The first is that the data was not confined to the United States. Public reporting on the Stolen Archive describes records covering people in the United States, the United Kingdom, and Canada. If you are listening to this in Kelowna or Calgary or Toronto and assuming that American broker exposure is an American problem, the record does not support you. The second is that essentially nobody had ever heard of national public data. It sold background checks to employers, landlords, and investigators. It had no consumer relationship with the people in its database. Nobody in that file agreed to be there. Nobody was told they were there. There was no account to close, and no setting to change, because there was no account. You cannot opt out of a company whose name you do not know, and there is no realistic way for you to learn the names of all of them. The third detail is the one I would ask you to remember. In October of that year, facing more than twenty class actions and inquiries from multiple state attorneys general, the parent company filed for bankruptcy. Court filings indicated it held less than seventy-five thousand dollars in assets and that its insurer had declined coverage, so the accountability went to zero. The people in that file received nothing. And then the following year, the site came back online with a search function. The company died, the data did not. That is the shape of this problem in a single sentence. You are not dealing with an organization which can be sued and can fail. You're dealing with a copy which cannot. At this point, the reasonable question is why the law does not simply fix this, and it is worth being precise about where that stands, because the direction of travel in the last two years has not been the one most people assume. In December of 2024, the Consumer Financial Protection Bureau in the United States proposed a rule that would have treated a wide range of data brokers as consumer reporting agencies under the Fair Credit Reporting Act. In practical terms, it would have forced them to meet accuracy obligations, to give people access to their own information, and to restrict who could buy sensitive categories of data. On the 15th of May, 2025, the Bureau withdrew it. The stated reason was that the rulemaking was not necessary or appropriate at this time. Whatever you think of that decision, the operational consequence is not ambiguous. There is no federal rule coming in the near term, and the burden of managing your own exposure stays exactly where it has always been, which is on you. In Canada, the position is different, but not more comfortable. Canadian private sector privacy law does not give you a general right to have your personal information deleted from a company that holds it. That is a common misconception, and it is worth correcting plainly. The obligation runs the other way. An organization is expected to stop retaining information once it no longer needs it for the purpose it collected it for. That is a duty on them, not a right you can exercise on demand. There has been movement. On the 27th of August, 2025, the Privacy Commissioner of Canada found that Google had contravened the federal statute, and that in limited circumstances an individual can require that certain results be removed from a search of their own name. The Commissioner framed the test carefully. It applies where there is a risk of serious harm, including harm to a person's safety or dignity, and where that risk outweighs the public interest in the information remaining findable. That is a meaningful development, and I do not want to understate it, but read what it actually grants, because the language is doing precise work. It grants delisting, it does not grant deletion. The commissioner was explicit that the material continues to exist on the sites that published it and remains reachable by other search terms. What changes is one index. The record is untouched. If someone knows where to look or knows to search for something other than your name, it is all still there. Delisting is not erasure, it is a curtain across one window. So where does that leave the premise we started with? It leaves it half right and the half that is right is worth having. You can reduce your exposure. That is real and it is worth doing, and doing it once with good research behind it will meaningfully shrink the number of places a casual searcher finds you in the first page of results. If you take nothing else from this episode, take that. Go and search your own name and your spouse's name and see what comes back with a home address attached to it. Most people listening have never done it, and the first time is instructive. But before you go further than that, there's a failure mode on the other side that almost nobody warns you about, and it is worth a minute of your attention because it runs against the instinct this whole subject creates. You can scrub too hard. Your ability to prove that you are you depends on a coherent history. Account recovery leans on it, identity verification at a bank leans on it, disputing a fraudulent account leans on it. Every one of those processes is underneath, a comparison between what you claim and what the record says. If you attack your own footprint aggressively and inconsistently, stripping some records, altering others, leaving a partial and contradictory trail behind you, then you degrade the thing those processes are checking against, and that creates a vacuum in a market where somebody else is holding a complete and internally consistent file on you. I want to be careful here, because this is not an argument against reducing exposure. It is an argument against confusing minimum footprint with maximum safety. Those are different objectives and they diverge quickly. The person who is hardest to impersonate is not the person with no record. It is the person with a coherent, controlled, verifiable record and a small number of high value details that are genuinely hard to find. Aim for controlled, do not aim for absent. You will not reach absent, and the attempt will cost you. Which brings us to the shift I actually want you to make, and it is the only thing I'm going to ask of you in this episode. Stop asking whether your data has been deleted. It has not been, and it will not be. Start asking a different question, which is whether you would know if something changed. Would you know if a new listing appeared under your name with your current address on it? Would you know if a broker that had honored your request last year quietly reacquired you this year? Would you know if your spouse's name started returning your home address because a property record was reindexed after a refinance? Would you know if a family member's post put a location on the house? Would you know if the exposure of a company you have never heard of put you in a file with two hundred million other people? A one time scrub cannot answer any of those questions. It answers one question, once, and then begins to expire immediately. What answers them is a standing capability that looks at the same surface repeatedly and tells you what is different this week from last week. And I want to give you a documented case of that working because it is easy to treat monitoring as an abstraction. Brian Krebs is an American journalist who writes about cybercrime. For roughly two decades, his work has involved publicly identifying people who commit fraud for a living, which means his name and his address have been known to hostile parties for most of his career. He has never had the option of deletion. That question was settled for him a long time ago. In March of 2013 he was one of the first journalists to be swatted. Somebody spoofed an emergency call claiming that Russians had shot his wife. He opened his front door and found a police officer aiming a handgun at him across the roof of a squad car. Separately, roughly $20,000 of credit was fraudulently taken out in his name. Then, later that year, a Ukrainian criminal operating under the handle fly organized a fundraiser on a carting forum. The plan was to buy heroin on a dark web market, have it shipped to Krebs' home in his name, and then spoof a call from one of his neighbors reporting him to the local police as a dealer so that the drugs would be found in his possession. The package never landed on him. Krebs had covert access to that forum. He watched the fundraiser assemble in real time, he read the plan as it was being made, and he notified law enforcement before the shipment arrived. The man behind it was later arrested in Italy, pled guilty in the United States, and was sentenced to forty one months. Sit with the structure of that, he could not delete himself, his exposure was total and permanent, and he had no realistic means of reducing it. What protected him was not a smaller footprint, it was that he was watching the place where the threat was forming, and he saw it before it reached him. That is the entire thesis of this episode in one case. The defense was not a razor, the defense was early sight. Now, most of you are not going to infiltrate a criminal forum, and you should not try. So let me be concrete about what the equivalent looks like at your level, because there is a second half to this that is entirely within your control and that most people skip. Exposure does not hurt you at the moment the data is collected, it hurts you at the moment somebody acts on a false instruction that the data made convincing. The wire that goes to the wrong account because the request came from a spoofed address and referenced a real transaction. The assistant who releases a schedule because the caller knew the name of the family lawyer, the bank that resets an account because the caller answered the security questions from a public record. So put your controls at that moment, not at the data. Four of them are worth having in place, regardless of anything else you do. A known good callback list, meaning a small set of verified numbers for the people who can move money or change your arrangements, and a standing rule that any instruction gets confirmed on a number from that list and never on a number supplied in the message itself. Two person approval on any change to payment instructions or banking details, with no exception for urgency because urgency is the pressure the attack is built on. One page hosted on infrastructure you control. That is the authoritative statement of who you are, who speaks for you, and how you can be verified. When somebody needs to check, they check there, and not against whatever a search returns, and a briefing for the people around you, your assistant, your spouse, your adult children, the person who manages the property. Most of the openings in a principal's life are not created by the principal. They are created by somebody adjacent who had no idea their post or their confirmation of your schedule was an intelligence contribution. Then recheck the exposure surface quarterly because the data will have repopulated and something will have changed. Which brings us back to New Orleans. Evan Ratliff did everything the deletion model tells you to do. He severed the records, he changed the name, the documents, the phone, the appearance, the location. By any audit of his paper trail he had disappeared, and he had done it competently enough to beat thousands of people for twenty five days. He was undone by a thing he could not delete, which was a pattern in how he had to live. Your position is not identical to his, but the lesson transfers cleanly. The records are the visible part and the part everybody focuses on, and they are worth reducing. But the durable exposure is not in any single record. It is in the shape of a life that continues to generate records, at addresses you keep returning to, on schedules you keep, in the names of people connected to you who never agreed to be careful. You can take down a listing, you cannot take down a pattern. So the honest goal is not to vanish. Nobody at your level of visibility is going to vanish, and anyone promising it is not being straight with you. The goal is to be difficult to assemble, hard to impersonate, and first to know when either of those changes. Deletion is a state you cannot hold. Monitoring is a posture you can take care of yourselves out there ages out.