The South Korean Bitcoin community appears to have suffered almost no direct losses from the recent Coldcard hardware-wallet attack, despite the devices being popular among experienced local holders.
The incident was linked to a flaw in the random-number generation used by certain Coldcard models, which reportedly exposed some seed phrases to attackers. Confirmed losses exceeded 1,596 BTC, worth roughly $130 million, while thousands of addresses may have been affected.
Coldcard manufacturer Coinkite responded by destroying remaining batches of vulnerable devices and urging users to generate new seed phrases. Independent estimates put the number of affected addresses at around 7,300, with potential losses reaching up to 2,000 BTC when suspected cases are included.
South Korea was a notable exception. According to Bitcoin analyst Koji Higashi, the relatively low losses were not simply a matter of luck. He pointed instead to a long-standing habit among Korean Bitcoin users of generating seed phrases independently rather than fully trusting a hardware wallet’s built-in randomness.
Korean Users Often Generated Seed Phrases Manually
For years, prominent members of the Korean Bitcoin community have encouraged users to generate their own entropy rather than rely entirely on the random-number generator built into a hardware wallet.
The methods are intentionally simple and physical:
- rolling dice or flipping coins to generate randomness;
- creating a BIP39 mnemonic completely offline;
- using 128 coin flips for a 12-word seed phrase or 256 flips for a 24-word phrase, then converting the binary result without relying on an internet-connected device.
Local guides explain the process step by step. Users can compare the resulting words with a printed BIP39 word list and use offline tools such as SeedSigner only to verify the checksum.
The idea is straightforward: the hardware wallet should not be the only source of randomness used to protect the funds.
Why English-Speaking Users May Have Been Hit Harder
English-speaking self-custody communities appear to have suffered considerably larger losses, despite generally having a high level of technical knowledge.
Higashi suggested that the difference may partly come down to how information and recommendations circulate within those communities.
One factor is the strong reliance on influential Bitcoin personalities. Some creators have sponsorship arrangements or close relationships with hardware-wallet manufacturers, which can make product recommendations appear more trustworthy than they otherwise would.
There is also an echo-chamber effect. When a product becomes widely accepted inside a technically sophisticated community, users may become less likely to question assumptions about its security.
The Korean community developed somewhat differently. According to Higashi, many of its most prominent educators had no commercial relationship with Coinkite and regularly advised users not to depend entirely on hardware-wallet firmware for seed generation.
In this case, many users appear to have followed that advice.
The broader lesson goes back to one of Bitcoin’s oldest principles: verify, don’t trust. That principle applies not only to software and hardware, but also to the people and sources recommending them.
For users managing significant amounts of Bitcoin, generating entropy independently can reduce reliance on a single device or manufacturer. The Coldcard incident is another reminder that even hardware wallets designed for advanced users should not automatically be treated as a completely trusted source of randomness.