Coldcard Hardware Wallet Flaw Linked to $70 Million Bitcoin Theft in 41 Minutes 10%

By Swati Khandelwal21%

8/1/2026, 10:17:00 AM

BS Summary: This article contains 18 faulty reasoning types, including Ambiguity (Equivocation), Anchoring Bias, and Attempt to Sell a Product or Service, with Availability Heuristic as the most egregious example at 18.3% saturation with 109 hits. Analysis detected 607 faulty-reasoning hits from 595 analyzed words, generating a BS Score of 21.2% and a BS Rank of 10% (24,372 of 26,881 articles). This article is better (less manipulative) than 90.70% of the article peer group.

An attacker drained 1,196 Bitcoin addresses in 41 minutes on July 30, taking 1,082.65 BTC worth about $70.2 million at the time. 
Galaxy Research mapped the sweep and tied it to a firmware flaw in Coldcard, the Bitcoin-only hardware wallet made by Canadian firm Coinkite. 
A March 2021 firmware integration error routed seed generation to a deterministic software pseudorandom number generator (PRNG) instead of the STM32 hardware random number generator (RNG). 
Block says an attacker who can determine or sufficiently constrain the device UID, timer state, and prior RNG-call history can reproduce candidate output streams offline without accessing the device. 
Candidate seeds can then be checked by deriving their addresses and comparing them with public blockchain data. 
Coinkite shipped emergency firmware for every affected model and release track on July 31, but installing it does not repair an existing seed. 
Coinkite tells owners with exposed seeds to generate a new one on patched firmware and move their coins. 
Restoring the old seed to updated firmware or another wallet carries the weakness forward. 
No public report has reconstructed a victim's seed and matched it to a drained address. 
Block traced the fault to Coldcard's production config, which defines MICROPY_HW_ENABLE_RNG as zero because Coinkite supplies its own hardware-RNG wrapper. 
The libngu library checked whether the macro existed rather than whether it was enabled, binding the build to MicroPython's Yasmarang fallback. 
The MicroPython fallback was initialized from the chip's unique ID and timer registers and collected no fresh entropy after initialization. 
Coinkite estimates effective entropy at roughly 40 bits on the Mk3 and about 72 bits on the Mk4, Mk5 and Q, against 128 bits for a 12-word BIP-39 seed. 
Block does not give one practical figure. 
It sets conditional ceilings below 240.7 and 273.3 and warns that the latter is not equivalent to 73-bit cryptographic security. 
It published no brute-force benchmark. 
The later-model reseed raises the number of candidates, but Block says practical cost depends on available UID information, boot timing, prior RNG calls and derivation cost. 
Exposure depends on the firmware running when the seed was created, not the version installed now: 
Mk2 and Mk3: Coinkite lists Mk3 versions 4.0.1 through 4.1.9, fixed in 4.2.0, and does not name Mk2. 
Block places both Mk2 and Mk3 versions 4.0.0 through 4.1.9 on the vulnerable path. 
Mk4 and Mk5: anything before 5.6.0. 
Q: anything before 1.5.0Q. 
Edge builds: before 6.6.0X for Mk4 and Mk5, before 6.6.0QX for Q. 
Coinkite says a seed built with at least 50 fair, independent, private dice rolls is not at risk from this bug alone. 
If the number or privacy of the rolls is uncertain, Coinkite says to migrate. 
A strong, unique BIP-39 passphrase creates a separate wallet the seed words cannot reach on their own, but the company still recommends replacing the seed. 
Multisig helps only when the quorum is not built entirely from affected devices. 
TAPSIGNER, OPENDIME and SATSCARD use different codebases and are unaffected. 
No one has named the attacker. 
Galaxy, which mapped the 1,196-address sweep, said it found no other Bitcoin transactions in the previous 30 days with the same 30 sat/vB, no-change signature. 
It warned that the pattern identifies the operator, not the theft, because a sweep "looks the same as if a coin owner chose to move coins." 
The disclosure follows Coinspect's Ill Bloom research in early July, a separate weak-PRNG flaw in older software wallets tied to more than $5 million drained from addresses across Bitcoin, Ethereum, Tron, Rootstock and Polygon since May. 
Article reasoning-pattern comparisonThis article: 2.4%Swati Khandelwal: 1.9%The Hacker News: 1.7%Confirmation Bias2.4%This article: 7.2%Swati Khandelwal: 1.1%The Hacker News: 1.0%Anchoring Bias7.2%This article: 18.3%Swati Khandelwal: 2.9%The Hacker News: 2.8%Availability Heuristic18.3%This article: 4.4%Swati Khandelwal: 1.1%The Hacker News: 1.2%Representativeness Heuristic4.4%This article: 0.0%Swati Khandelwal: 0.5%The Hacker News: 0.5%Hindsight Bias0.0%This article: 4.9%Swati Khandelwal: 1.9%The Hacker News: 2.2%Overconfidence Bias4.9%This article: 0.0%Swati Khandelwal: 2.0%The Hacker News: 2.2%Framing Effect0.0%This article: 0.0%Swati Khandelwal: 0.6%The Hacker News: 0.8%Loss Aversion0.0%This article: 3.0%Swati Khandelwal: 0.5%The Hacker News: 0.5%Status Quo Bias3.0%This article: 0.0%Swati Khandelwal: 0.1%The Hacker News: 0.1%Sunk Cost Effect0.0%This article: 4.2%Swati Khandelwal: 0.9%The Hacker News: 1.1%Optimism Bias4.2%This article: 0.0%Swati Khandelwal: 1.5%The Hacker News: 1.3%Pessimism Bias0.0%This article: 2.5%Swati Khandelwal: 4.7%The Hacker News: 5.4%Negativity Bias2.5%This article: 3.7%Swati Khandelwal: 0.3%The Hacker News: 0.6%Self-Serving Bias3.7%This article: 0.0%Swati Khandelwal: 0.3%The Hacker News: 0.3%Fundamental Attribution Error0.0%This article: 0.0%Swati Khandelwal: 0.1%The Hacker News: 0.0%Actor-Observer Bias0.0%This article: 1.7%Swati Khandelwal: 0.1%The Hacker News: 0.1%In-Group Bias1.7%This article: 0.0%Swati Khandelwal: 0.1%The Hacker News: 0.3%Out-Group Homogeneity Bias0.0%This article: 0.0%Swati Khandelwal: 0.2%The Hacker News: 0.4%Halo Effect0.0%This article: 0.0%Swati Khandelwal: 0.0%The Hacker News: 0.0%Horn Effect0.0%This article: 0.0%Swati Khandelwal: 0.0%The Hacker News: 0.0%Dunning-Kruger Effect0.0%This article: 6.1%Swati Khandelwal: 1.2%The Hacker News: 1.3%Recency Bias6.1%This article: 0.0%Swati Khandelwal: 0.2%The Hacker News: 0.2%Primacy Effect0.0%This article: 0.0%Swati Khandelwal: 0.1%The Hacker News: 0.1%Blind-Spot Bias0.0%This article: 0.0%Swati Khandelwal: 0.0%The Hacker News: 0.1%Ad Hominem0.0%This article: 0.0%Swati Khandelwal: 0.0%The Hacker News: 0.1%Straw Man0.0%This article: 3.9%Swati Khandelwal: 3.1%The Hacker News: 3.2%Appeal to Authority3.9%This article: 0.0%Swati Khandelwal: 0.9%The Hacker News: 1.2%False Dilemma0.0%This article: 0.0%Swati Khandelwal: 0.5%The Hacker News: 0.5%Slippery Slope0.0%This article: 0.0%Swati Khandelwal: 0.0%The Hacker News: 0.0%Circular Reasoning0.0%This article: 0.0%Swati Khandelwal: 2.8%The Hacker News: 3.6%Hasty Generalization0.0%This article: 0.0%Swati Khandelwal: 0.1%The Hacker News: 0.1%Red Herring0.0%This article: 0.0%Swati Khandelwal: 0.1%The Hacker News: 0.1%Bandwagon0.0%This article: 0.0%Swati Khandelwal: 0.6%The Hacker News: 0.9%Appeal to Emotion0.0%This article: 0.0%Swati Khandelwal: 0.2%The Hacker News: 0.4%Begging the Question0.0%This article: 6.1%Swati Khandelwal: 1.4%The Hacker News: 1.6%Post Hoc (False Cause)6.1%This article: 0.0%Swati Khandelwal: 0.0%The Hacker News: 0.0%Tu Quoque0.0%This article: 0.0%Swati Khandelwal: 0.7%The Hacker News: 0.5%Burden of Proof0.0%This article: 0.0%Swati Khandelwal: 0.1%The Hacker News: 0.1%Appeal to Nature0.0%This article: 0.0%Swati Khandelwal: 0.2%The Hacker News: 0.3%Composition/Division0.0%This article: 0.0%Swati Khandelwal: 0.8%The Hacker News: 0.8%Anecdotal0.0%This article: 0.0%Swati Khandelwal: 0.1%The Hacker News: 0.1%No True Scotsman0.0%This article: 15.3%Swati Khandelwal: 1.9%The Hacker News: 1.8%Ambiguity (Equivocation)15.3%This article: 0.0%Swati Khandelwal: 0.0%The Hacker News: 0.0%Gambler’s Fallacy0.0%This article: 0.0%Swati Khandelwal: 0.0%The Hacker News: 0.0%Middle Ground0.0%This article: 0.0%Swati Khandelwal: 0.0%The Hacker News: 0.0%Personal Incredulity0.0%This article: 3.9%Swati Khandelwal: 0.1%The Hacker News: 0.1%Special Pleading3.9%This article: 0.0%Swati Khandelwal: 0.1%The Hacker News: 0.1%Genetic Fallacy0.0%This article: 0.0%Swati Khandelwal: 0.9%The Hacker News: 1.2%Unattributed Quote0.0%This article: 0.0%Swati Khandelwal: 0.5%The Hacker News: 0.8%Quote-first Misdirection0.0%This article: 3.7%Swati Khandelwal: 1.9%The Hacker News: 1.9%Biased Writer Voice3.7%This article: 4.2%Swati Khandelwal: 3.6%The Hacker News: 3.3%Indoctrination4.2%This article: 0.0%Swati Khandelwal: 0.0%The Hacker News: 0.0%Politically Left Leaning Bias0.0%This article: 0.0%Swati Khandelwal: 0.0%The Hacker News: 0.0%Politically Right Leaning Bias0.0%This article: 6.7%Swati Khandelwal: 0.5%The Hacker News: 2.5%Attempt to Sell a Product or S…6.7%

595 words analyzed.

Speakers

1speaker13%attributed speech516writer words
Selected voice

Coinkite

100%flagged-word coverage
79 attributed words100% of attributed speech61% writer coverage
0%27.5%55.0%Attempt to Sell a Product +50.6 ptsWriter: 0.0%Coinkite: 50.6%50.6%Indoctrination+31.6 ptsWriter: 0.0%Coinkite: 31.6%31.6%Biased Writer Voice-4.3 ptsWriter: 4.3%Coinkite: 0.0%0.0%

Attribution is sentence-level. Pattern percentages are calculated only from words assigned to that voice.

Loading…
Loading…
Loading…
Loading…

Analysis

Hover over highlighted words in the article to view the associated bias or fallacy analysis.