Belgian Researchers Put Historic Bitcoin (BTC) Exposure at $106 Million
University of Liège researchers traced a nonce-reuse flaw exposing about $106 million in Bitcoin (BTC) funds, with bots striking bait wallets within seconds.
AI SummaryAI
- University of Liège researchers recovered 3,278 Bitcoin (BTC) private keys in a six-chain scan through June 30, 2024.
- Automated bots attacked deliberately vulnerable bait wallets within seconds, with at least two rival transactions competing.
- Peak historical exposure of vulnerable addresses totaled about $106 million at Nov. 25, 2024 prices.
- A bait address's remaining 5,200 satoshis were swept in a transaction confirmed nine minutes later.
Bots Strike Bait Wallets in Seconds
Automated thieves hit deliberately vulnerable
Bitcoin (BTC) wallets within seconds after two Belgian researchers wired up the flaw on purpose. Vincent Jacquot and Benoît Donnet, computer scientists at the University of Liège's Montefiore Institute, loaded $3 to $5 each into Bitcoin and Ethereum accounts left open to a signing error capable of revealing private keys. On the Bitcoin side, the attack began a second or a few seconds after the bait went live, Jacquot reported this week. At least two competing transactions raced for the coins inside the mempool, the public holding area where pending transfers wait before the network confirms them. The mistake sits in how wallets sign payments. Every signature relies on a nonce, a random number meant to be used exactly once; if a secret key reuses one, anyone who sees both signatures can compute the key with simple algebra. Jacquot and Donnet sent 7,200 satoshis, bitcoin's smallest unit and worth a few dollars, to each of six newly created addresses from a single wallet. About an hour later, two of those addresses paid each other, both signing with the same nonce; that alone exposed nothing, because two different keys did the signing. The break came when one of the pair signed a third payment with that nonce, so a single key had reused it and the unlocking math was public. The next payment out of the bait sent its remaining 5,200 satoshis to a stranger in an on-chain sweep signed with the bait's own private key. A block confirmed it nine minutes later, and the 4,700 satoshis in the thief's wallet have not moved since. The partner address fell by extension, because the shared nonce surrendered that key as well, and it sat untouched for 13 days before another sweep cleared it months later.
3,620 Keys, $106 Million in Exposure
The bait experiment scaled into an audit. Under the most common signing method, ECDSA, wallet software builds each signature from two ingredients: the owner's private key and a fresh random number meant for single use. The pair went through all ECDSA signatures across six blockchains, from each chain's first block up to June 30, 2024, hunting for the reuse mistake. They recovered 3,620 private keys in all: 3,278 on
Bitcoin (BTC), 195 on Dogecoin, 119 on Ethereum, nine on Litecoin, six on Bitcoin Cash and one on Dash. Fifteen keys turned up only when all six networks were searched together, showing that some faulty wallet software had reused nonces on different chains; the addresses behind those keys peaked at just under $2,100 combined. Each vulnerable address's highest recorded balance was valued at cryptocurrency prices from Nov. 25, 2024, and the peaks summed to about $106 million, or €101 million. The estimate assumes an attacker caught every address at its richest moment and that victims never topped up afterward, and the team did not establish how much money was actually stolen. Most of the exposure sits in bitcoin's early years. The exposed pot crossed $1 million nine times, with most of those moments between 2014 and early 2016; the record was about $26 million in December 2014. Another spike of roughly $12 million arrived in 2022 and about $1 million followed in 2023, both mostly from Ethereum, where 6 ETH and then 350 ETH entered one exposed address in November 2022 and moved to a different address 12 seconds after arriving. Attackers appear selective, too: the authors judge Dogecoin, Dash and Litecoin unlikely to face active scrutiny, and most currently vulnerable funds now sit in a single
Bitcoin (BTC) address with an uncommon format, which the paper does not publish. The earliest activity flagged as a possible attack dates to the fourth quarter of 2015.
The on-chain record carries the weight here: one of the trap addresses shows the sweep signed by the bait's own key minutes after the nonce was reused. What the study cannot show is a theft total. The $106 million is an exposure ceiling priced at Nov. 25, 2024, and the authors themselves do not separate attacks from ordinary withdrawals. The structural point: once a pending transaction sits in the mempool, everyone can see it, so predictable errors become standing opportunities for bots. Wallet vendors of every kind, from an HD wallet to a Tangem Mobile Wallet, carry the duty of keeping the one-time rule from breaking. July's Coldcard incident, where a key-generation flaw allowed thieves to pull 594 BTC out in an initial 25-minute wave, showed the same. Until those gaps close, DYOR has to extend to the math inside the wallet.
AI-generated, AI-reviewed, under COINOTAG editorial oversight.

