23andMe Sold the Map of You. Twice.

You did not send them a tube of spit. You sent them a map.

A map of you — not where you live, but what you are made of, and who you are related to. You clicked through the terms of service the way everyone does, late at night, half reading, trusting the brand name on the box. The kit arrived, you did the thing, you mailed it back, and you forgot about it. That was the deal, as you understood it: a private curiosity, answered once, kept by a respectable company.

Then the company was breached. Then the company went bust. And the map you handed over became an asset on someone else's balance sheet.

That is the story in one sentence, and it is worth sitting with, because it is not really a story about one company. It is a story about what happens when the most unseeable thing about you is entrusted to something that can fail.

First, it was stolen

In October 2023, 23andMe suffered a data breach that was widely reported. User data — the kind of data that describes people, not just accounts — was exposed. For most breaches, the advice is familiar and almost comforting: change your password, turn on two-factor authentication, watch your statements. It is the standard script, and it works, because most of what leaks can be replaced.

Genetic data cannot be replaced. You can change a password. You cannot change your genome. You cannot change your relatives' genomes either, which is the part people tend to forget until it is too late: a map of you is, in part, a map of your family. When it leaks, it does not leak about one person. It leaks about a bloodline.

That alone should have changed how we think about handing this kind of material to an online account. It did not, because the company was still standing, still sending cheerful emails, still behaving like a permanent institution. We are very good at treating companies as if they were geology — solid, immovable, always there. They are not. They are businesses. And businesses have fates.

Then, it was for sale

In March 2025, 23andMe filed for bankruptcy protection. The news was reported everywhere, and the reaction was immediate and telling: people did not ask about the share price. They asked about their data. What happens to the map now? Who buys it? What can the buyer do with it?

Those are the right questions, and they are also unanswerable in any way that should reassure you — because the answers depend on a bankruptcy court, a buyer, and a new set of terms of service written by people you never chose. The company you trusted is gone. The asset remains. And an asset, by definition, goes to whoever pays for it.

This is the second sale, and it is quieter than the first but more instructive. The first time, the map was taken by intruders. The second time, it was offered up by the ordinary machinery of business: a company in trouble, disposing of what it owns. Nobody had to break in. The front door was opened by a filing.

The illusion: "a big company will take care of it"

Most of us outsource our worrying to brand names. We assume that a company large enough to advertise on television, large enough to be a household name, must have solved the problem of permanence. It has not. It has only solved the problem of scale — how to hold a great many people's things at once, in one place, under one set of rules that can be rewritten.

An online account is a single point of failure wearing a friendly interface. The terms of service can change by a page. The ownership can change by a signature. The servers can change hands without you ever being consulted, because you were never the customer in the sense that matters: you were the source of the asset. When the company fails, your data does not evaporate out of politeness. It is inventoried, valued, and transferred.

This is not an argument against technology. It is an argument against a category error: treating a corporation's fortunes as a safe place to keep something irreplaceable. A company's fate is, by design, uncertain. Your privacy cannot rest on something uncertain and still be called private.

What certainty actually looks like

Certainty is boring, and that is the point. Certainty is knowing, before you hand something over, exactly whose hands it will be in, whether those hands will change, and who can see it. It is not a promise in a terms-of-service document that a future owner can revise. It is an arrangement nailed down in advance: who holds it, under what seal, under what rules, and what happens in the worst case — spelled out in a contract and a trigger ledger, not left to a bankruptcy proceeding to decide.

That is why the answer is not a better app, or a bigger company, or a stronger password. The answer is a person and a process you can hold to account. A numbered item, sealed, photographed at intake, checked by two people rather than one, kept by professionals whose entire job is custody — not growth, not advertising, not an exit. The technology underneath matters, of course; encryption is the floor, not the ceiling. But the thing that makes it hold over years is not a feature. It is the discipline of the people and the paper trail around it.

Consider the difference in the worst case. If a cloud company fails, your data is an asset to be sold. If a custody arrangement fails, there is a contract that already says what happens, a ledger that already records every trigger, and a sealed item that was never sitting on a server waiting to be transferred. One worst case is decided by strangers after the fact. The other is decided by you, beforehand.

The question to ask before you hand anything over

Here is the test I would apply to anything you would not want found — not just genetic data, but anything that belongs in the category of "never to be seen." Before you entrust it, ask: if this company disappears tomorrow, where does my material go, and who decides? If the honest answer involves a court, a buyer, or a revised set of terms you have not read, you do not have certainty. You have a hope, dressed up as a service.

And ask a second question: can what I am handing over be taken back? A password can. A photo can, mostly. A genome cannot. The less reversible the material, the less it belongs anywhere that can change owners, change policies, or change its mind. Irreversible things demand irreversible arrangements — sealed, offline, out of any system that treats them as inventory.

Seneca wrote that we suffer more in imagination than in reality. With genetic data, the reverse is true: the reality is worse than most people imagine, because they imagine a breach as the worst case. The breach was only the first sale. The second sale — the legal, orderly, entirely unsurprising transfer of your map to a stranger — is the one that should change how you choose where to keep anything you cannot afford to have seen.

Do not hand the map of you to a fate. Hand it to a person, under a seal, under a contract, where no balance sheet can sell it twice.