Bitcoin (BTC) Coldcard Flaw Drains 1,082 BTC

A Coldcard firmware flaw let attackers drain 1,082.65 BTC in 41 minutes, while COINOTAG data shows Bitcoin near $63K with strong support at $62.4K.

(05:40 AM UTC)
6 min read
Updated
AI SummaryAI
  • A Coldcard firmware flaw drained 1,082.65 BTC from 1,196 addresses on July 30 in a 41-minute window.
  • Blockchain analytics found about $30 million extracted in the first 10 minutes, with the initial 25-minute sample exceeding $38 million.
  • One early target held around $1.8 million, and three of the 10 most-affected wallets contained more than 10 BTC each.
  • The exploit reduced seed entropy from 128 bits to roughly 40 to 72 bits on vulnerable 2021 firmware.
LDR

A security failure in a long-used Bitcoin hardware wallet drained 1,082.65 BTC from 1,196 addresses on July 30, according to on-chain analysis reviewed by COINOTAG, making it one of the largest self-custody breaches tied to a single device family. The transfers occurred in a 41-minute window and used an identical transaction fee, a pattern that points to automated control of compromised private keys rather than manual user activity. The affected devices include Coinkite's legacy Coldcard Mk3 and some units running older firmware on Mk4, Mk5 and Q models, where seed generation may have been weakened. Stolen coins were swept into four clustering addresses, and the victim set appears concentrated among individual holders rather than exchanges or institutions. The incident raises difficult questions for the Bitcoin custody model, because wallets set up years ago may remain unmonitored until funds are already gone. Some users reported retrieving old recovery phrases just in time, but others may not discover losses until a future all-time high prompts them to check balances. The episode also revived debate over whether dormant long-term holders should rely more on regulated custodial products instead of purely self-managed keys. For Bitcoin holders, the practical lesson is to verify firmware history, move vulnerable seeds to freshly generated addresses, and treat cold storage as an actively maintained security process.

Blockchain analytics of the same Coldcard exploit show a deliberate sequencing pattern, with the attacker prioritising the largest balances before moving down the list. The review found that about $30 million was extracted within the first 10 minutes, and the initial 25-minute sample exceeded $38 million across roughly 500 wallets. One early target held around $1.8 million, while three of the 10 most-affected wallets contained more than 10 BTC each, indicating that the campaign was likely prepared in advance rather than opportunistic. Coinkite's advisory identified vulnerable Mk3 firmware versions spanning 4.0.1 to 4.1.9 and warned that installing a hotfix alone does not rescue an already weak seed. Users who generated recovery phrases on affected firmware are being urged to create a new seed on patched hardware and add a strong BIP-39 passphrase, which can raise the effective cost of brute-force attacks. The stolen funds were still being monitored for movement, and researchers said related addresses may remain exposed to automated sweeps. This makes wallet hygiene a continuing operational task, not a one-time setup. For self-custody users, the checklist includes verifying device firmware dates, rotating any seed created during the vulnerable period, and keeping offline recovery backups separate from networked machines.

The technical cause behind the Coldcard losses centers on weak entropy during seed creation, according to the deeper on-chain and firmware review. Certain older firmware releases from 2021, including versions around 4.0.0 and 4.0.1, allegedly bypassed the device's hardware random-number generator and relied on a more predictable software path. That reduced effective seed entropy from the expected 128 bits to roughly 40 to 72 bits, making private-key derivation far easier for an attacker with sufficient compute. The swept addresses were overwhelmingly Native SegWit accounts, with 1,183 of the 1,196 drained addresses using the BIP-84 format. Although most emptied wallets held less than 1 BTC, the economic damage was concentrated in balances of 1 to 50 BTC, a profile consistent with individual self-custody rather than exchange wallets. Each malicious transaction paid a hardcoded 30 sat/vB fee, far above the weekly median, and carried no change output, another signature of automated spending. The activity spanned six blocks, with three blocks showing no sweeps, suggesting batched broadcasting. The sweep also began about 30 hours before the public security warning, leaving unaware users exposed during that interval. Coinkite has told users that adding strong extra entropy, such as many dice rolls or a robust BIP-39 passphrase, can reduce exposure, but migrating to a newly generated seed remains the safest remedy.

The total loss has since been revised upward to approximately $70 million, nearly double the $38 million figure cited in early reports. Changpeng Zhao, founder and former CEO of Binance, publicly warned that even long-trusted hardware wallets can contain latent bugs, advising holders to distribute funds across multiple devices while acknowledging that approach introduces its own trade-offs. Separately, Block's engineering team identified the operator through a paid subscription at a blockchain-data provider and has referred its findings to law enforcement. Block also flagged the older Coldcard Mk2 as potentially vulnerable, a model absent from Coinkite's own advisory. As of the latest on-chain data, the stolen coins remain unmoved in the four clustering addresses, though researchers caution that additional sweeps are possible while weak seeds still hold balances.

The root cause has been traced to a specific code-level failure: an internal build flag disabled the hardware random-number generator, and a supporting library verified only the presence of that flag rather than confirming it was active, allowing seed derivation to default to a software substitute keyed from the chip's factory serial number and clock registers. For the Mk4, Q and Mk5, researchers placed the resulting enumerable keyspace at roughly four billion candidates — trivially searchable with commodity hardware. Galaxy Research further noted that drained wallets spanned three distinct address formats, consistent with a blanket enumeration scan rather than targeted theft. No self-diagnostic exists that lets an owner confirm whether their seed falls inside the compromised range, so anyone who generated a key on affected firmware must treat their holdings as exposed and migrate immediately.

(as of 20:56 UTC) COINOTAG's proprietary 42-indicator composite S/R scoring engine shows Bitcoin (BTC), as of the latest snapshot, trading near $62,651 after a 0.45% daily decline, with the $63,161 resistance rated 70/100 from Fibo 0.214, HVN, SMA 50, and Pivot Point confluence. The strongest support at $62,309 scores 83/100, driven by Ichimoku Senkou A, Ichimoku cloud-bottom, Donchian lower, and Swing Low signals. Derivatives positioning remains long-leaning with funding marginally positive at 0.0029%, open interest sits at $12.7 billion, and the long/short account ratio is 2.22, while Fear and Greed reads 27, or Fear. A reclaim of $63,161 could open a test of $66,503, but losing $62,309 would confirm the downtrend and expose $61,044. Bitcoin's 69.7% tracked-market share keeps the altcoin complex vulnerable in any renewed bear market leg.

728
Sarah Chen

Sarah Chen

COINOTAG author

View all posts
AI-AssistedMarket Analyst·Sarah Chen is a market analyst specializing in technical analysis and risk management for cryptocurrency markets, with five years of active trading desk experience.

AI-generated, AI-reviewed, under COINOTAG editorial oversight.