Key Highlights
- University of Liège researchers revealed that automated mempool bots swept funds from deliberately vulnerable Bitcoin bait wallets within mere seconds of a cryptographic signing error.
- A multi-chain audit covering ECDSA signatures up to June 30, 2024, recovered 3,620 private keys across six blockchains, including Bitcoin, Ethereum, and Dogecoin.
- The theoretical peak exposure of the compromised addresses reached approximately $106 million (€101 million) based on November 25, 2024 cryptocurrency market valuations.
Automated Bots Drain Flawed Crypto Wallets Within Seconds
In a controlled cybersecurity experiment, automated scanning bots detected and drained intentionally flawed Bitcoin addresses within seconds of exposure. Computer scientists Vincent Jacquot and Benoît Donnet from the Montefiore Institute at the University of Liège set up bait wallets loaded with $3 to $5 worth of Bitcoin and Ethereum to evaluate how aggressively threat actors monitor public ledgers for exploitable cryptographic vulnerabilities. On the Bitcoin network, the exploitation began just a second or a few seconds after the vulnerable transaction broadcasted. Vincent Jacquot noted that at least two competing transactions immediately entered the mempool—the decentralized staging area where unconfirmed transactions wait—competing in an automated race to siphon the funds.
The bait architecture relied on a specific mathematical oversight involving digital signatures. To execute the test, Jacquot and Donnet distributed 7,200 satoshis to six newly generated addresses originating from a single wallet. Roughly an hour later, two of these addresses executed transfers to one another using an identical nonce. While using the same nonce across two distinct private keys did not compromise either, a subsequent third transaction by one of the addresses reused that same nonce. Once a single private key signed multiple transactions with identical nonces, the mathematical barrier collapsed, allowing an external observer to compute the underlying private key using basic algebra. An automated thief promptly generated an on-chain transaction signed with the compromised key, sweeping 5,200 satoshis. The sweep was confirmed in a block nine minutes later, leaving 4,700 satoshis in the unauthorized wallet, where the funds have remained unmoved. The interconnected partner address fell by extension due to the shared nonce, remaining dormant for 13 days before being cleared in another transaction months later.
How ECDSA Nonce Reuse Exposes Private Keys
The vulnerability exploited by these monitoring bots stems from implementation errors within the Elliptic Curve Digital Signature Algorithm (ECDSA), the primary cryptographic standard utilized by legacy cryptocurrency networks. Under ECDSA, an authorized signature requires two inputs: the user’s secret private key and a unique, unpredictable random number known as a nonce. The protocol demands that every nonce be used strictly once per signature. If flawed wallet software fails to generate cryptographically secure random values and repeats a nonce across two different signatures generated by the same private key, the key can be extracted instantly through elementary algebraic calculations.
Large-Scale Blockchain Audit Uncovers $106 Million in Cumulative Risk
Following the bait experiment, the University of Liège researchers scaled their methodology into an expansive historical audit. The investigation examined every ECDSA signature published across six major networks—Bitcoin, Ethereum, Dogecoin, Litecoin, Bitcoin Cash, and Dash—from their respective genesis blocks up to June 30, 2024. Through this comprehensive sweep, the researchers recovered a total of 3,620 unique private keys. The vast majority of the compromised credentials, 3,278 keys, were discovered on the Bitcoin blockchain. The remaining discoveries included 195 keys on Dogecoin, 119 on Ethereum, nine on Litecoin, six on Bitcoin Cash, and one on Dash.
The researchers also identified 15 private keys that were only exposed by cross-referencing datasets across all six networks simultaneously, revealing that certain faulty wallet implementations had repeated nonces across different blockchains. The addresses tied to these cross-chain failures reached a combined historical high of just under $2,100. When calculating the highest recorded balances of all 3,620 compromised addresses using asset prices recorded on November 25, 2024, the aggregate peak theoretical exposure stood at approximately $106 million (around €101 million). This metric assumes each vulnerable wallet was targeted at its historical peak balance and excludes subsequent top-ups, though the researchers did not determine the total sum of cryptocurrency successfully stolen by malicious actors over the period.
Historical Trends and Attack Behavior Across Major Chains
The audit demonstrated that the majority of significant nonce reuse incidents occurred during the early developmental phases of the cryptocurrency ecosystem. Aggregate exposed balances crossed the $1 million threshold on nine separate occasions, with the bulk of those events concentrated between 2014 and early 2016. The peak historical exposure reached roughly $26 million in December 2014. However, modern vulnerabilities have also emerged; a notable spike of approximately $12 million appeared in 2022, followed by another $1 million exposure in 2023, both centered heavily on Ethereum. In one documented November 2022 incident, deposits of 6 ETH and 350 ETH entered an exposed Ethereum address, only to be swept to an external address just 12 seconds later.
Active surveillance by threat actors appears highly targeted by chain. According to Jacquot and Donnet, networks such as Dogecoin, Dash, and Litecoin do not appear to face the same relentless, real-time mempool scanning seen on Bitcoin and Ethereum. Additionally, the researchers found that the majority of currently vulnerable capital is preserved in a single Bitcoin address utilizing an uncommon script format, which the authors deliberately withheld from publication. The earliest historical on-chain trace indicating automated predatory exploitation dates back to the fourth quarter of 2015.
Why This Matters
The findings from the University of Liège emphasize the relentless and automated nature of modern blockchain surveillance. In public, transparent mempools, autonomous bots continuously monitor live network traffic for structural weaknesses, turning implementation mistakes into instantaneous losses. While modern cryptographic libraries frequently implement deterministic nonce generation (such as RFC 6979) to mitigate accidental nonce reuse, legacy wallet architectures, flawed random number generators, and cross-chain operational oversights continue to present serious risks. The study demonstrates that even brief or obscure mathematical flaws on public blockchains are actively hunted by actors capable of executing exploits before transactions even achieve their first network confirmation.
Frequently Asked Questions
What is an ECDSA nonce reuse vulnerability?
Under the ECDSA cryptographic standard, every transaction signature requires a secret private key and a freshly generated random number known as a nonce, designed to be used only once. If a software wallet reuses the same nonce to sign two different transactions using the same private key, an outside observer can derive the private key directly from the public blockchain using simple algebraic formulas.
How quickly do automated bots steal funds from vulnerable wallets?
In the controlled study conducted by researchers Vincent Jacquot and Benoît Donnet, predatory mempool bots detected the nonce flaw and broadcast competing transactions to sweep the funds within a single second to a few seconds after the vulnerable transaction appeared on the Bitcoin network.
Which cryptocurrency networks were examined in the research?
The researchers audited ECDSA signatures from network inception through June 30, 2024, across six prominent blockchains: Bitcoin, Ethereum, Dogecoin, Litecoin, Bitcoin Cash, and Dash, successfully recovering 3,620 private keys across the analyzed ledgers.




