Back to Exposure Report
18,835 devices to be destroyed October 20, 2026 • Cryptocurrency Hardware

Keystone Hardware Wallet

The order record says who bought a device for storing cryptocurrency, and where to find them.

NamesContact detailsShipping addressesOrder detailsDevice authentication key
1

What happened?

Keystone, a hardware wallet manufacturer, detected unauthorised access to its official website on 20 October 2026. On 23 October the company confirmed, after reviewing API activity with Shopify, that an unidentified third party had obtained customer information. Keystone publicly disclosed the incident on 24 October.

On 26 October the disclosure widened considerably. Keystone said the attacker had also obtained the authentication key its hardware wallets use to prove they are genuine. The company stopped using the key, suspended shipments of devices carrying it, and said it will destroy 18,835 units.

2

What data was actually inside?

Names, contact details, delivery addresses and order details including the items purchased and the prices paid. Keystone states that payment card information, bank details, wallet recovery phrases, private keys and passwords were not compromised.

Separately, the device authentication key. That is not customer data at all. It is the secret that underwrites the supply chain integrity check a buyer performs to confirm a device has not been tampered with.

3

Who gets hurt and how?

Buyers of hardware wallets, who are self selected as people holding cryptocurrency in amounts large enough to justify dedicated hardware, and whose home addresses are now in a list that says so.

This is the rare breach where the primary risk is physical. The combination of a name, a residential address and a purchase record for a cold storage device identifies a household that probably holds bearer assets on the premises, and the price paid is a crude proxy for how seriously. Cryptocurrency theft by coercion is a documented pattern, and the hard part for an attacker has always been target selection. This list solves target selection.

The device key is a second, slower harm. Anyone holding it could build counterfeit units that pass the online genuineness check an owner uses to verify a wallet, which inverts the exact control that was supposed to protect buyers from a tampered device.

4

What did they think they were doing right?

The crypto assets were defended and the defence held. Recovery phrases, private keys and passwords are the material that matters for a wallet company, and none of it was taken. Keystone also ran the genuineness check that responsible hardware vendors are expected to run.

The gap is that the threat model stopped at the asset. A hardware wallet company reasonably spends its security budget on the seed phrase, because that is what converts directly to money. The order table was an ecommerce artifact sitting in Shopify, classified the way every ecommerce order table is classified, with no adjustment for the fact that the product being ordered redefines what the address means.

5

What did they not know about their own data?

That the order table was the most dangerous database in the company. Names and shipping addresses are ordinary retail fields and they are scored as ordinary retail fields by every classification scheme in use. What changes the score is the product, and the product is not in the sensitivity calculation.

This is the same failure of context seen in the Daishin Securities entry in this series, where three low sensitivity fields became a target list because of who they described. Here the fields become a target list because of what the people bought.

The device key is a different and sharper version of the problem. It was a secret whose compromise invalidates physical inventory, and the recovery cost is now measured in 18,835 units being destroyed rather than in notification letters.

If your business runs on databases, you probably have similar records, customer data, credentials, financial information. Do you know what's actually in yours?

Not sure where to start? How to find sensitive data across your estate

6

What does attribution look like the morning after?

No threat actor has been named. The timeline is unusually well documented: access detected 20 October, scope confirmed with the platform provider on 23 October, public notice 24 October, and the key compromise disclosed on 26 October.

That escalating sequence is the honest shape of an investigation and it is also the hardest thing to communicate to customers. Each update expanded the blast radius, and a buyer reading the 24 October notice had no way to know that the 26 October notice was coming. The operational burden here is not notification volume. It is telling people whose risk is physical that the picture has changed again.

7

What would have changed the outcome?

Scoring the order table by what it implies rather than by which columns it contains. A name and an address attached to a cold storage purchase is not retail data and should never have been governed as retail data.

Second, minimising it. Shipping addresses have an operational life measured in weeks. Order history for a product like this has an operational life of approximately zero once the warranty window closes, and every month it is retained is a month it can be stolen.

Keystone Hardware Wallet found out the hard way.

Your team could spend the next 6 months rebuilding systems, notifying customers, and answering legal questions. Or you could spend 24 hours finding out what's actually at risk.

Want to understand the process first? Read our guide to sensitive data discovery .