A study presented at USENIX Security '26 identified 65,340 risky crypto addresses involved in misuse across Ethereum and BNB Smart Chain, with 126,982.94 ETH and 17,726.7 BNB in associated native-token losses. The researchers valued those losses at more than $574.8 million. But two newly described active attack vectors directly account for about $15.7 million, or 2.7% of that figure. The full paper also used May 2025 reference prices of $4,408 per ETH and $847 per BNB, rather than each token's dollar value when the losses occurred.
That price discrepancy is the first structural flaw. Using current prices for historical losses inflates the headline figure. More importantly, the separation between detected misuse and active attack vectors reveals a fundamental problem in how we quantify risk. The 65,340 addresses are a rogues' gallery of exposed private keys, misrouted contract calls, and abandoned code. The $574 million is a liability estimate, not a direct loss. The two active vectors represent only 3,472.23 ETH and 465.24 BNB โ a fraction of the total. Yet these are the vectors that demand immediate remediation.
Context: The Anatomy of Misuse
The study separates misuse into two categories: contract-account misuse and externally owned account (EOA) misuse. Contract-account misuse occurs when a user sends a function call, sometimes with ETH or BNB attached, to an address that has no contract code on the selected network. The transaction can still succeed as a simple transfer without executing the intended function. Funds then sit at that address unless later-deployed code can move them. This is not a bug; it is a feature of Ethereum's account model. But it becomes a trap when attackers exploit deterministic contract addressing.
The second category, EOA misuse, starts with a public or otherwise exposed private key. Anyone who has the key can control the account, and automated sweepers can race to remove incoming funds. The study extracted over 16.3 million deduplicated private keys from GitHub, derived their addresses, and combined direct key matches with transaction-pattern rules and lightweight symbolic execution. The precision rate for the detection results is reported at 99.11%. That is impressive, but precision measures detection accuracy, not the direct causation of losses.
Core: The Two Active Vectors โ A Systematic Teardown
The first active vector exploits deterministic contract addressing. An attacker deploys a contract at a testnet address, waits for users to mistakenly send funds to the corresponding no-code address on mainnet, then deploys malicious withdrawal code at the same location. The paper identified 469 malicious contracts tied to 3,446.37 ETH and 431.79 BNB in losses. This is a classic example of what I call a "structural inefficiency" โ the protocol's deterministic address generation becomes a liability when combined with cross-chain carelessness.
Based on my audit of the Geth client in 2017, I identified a similar race condition in transaction propagation. The Geth team fixed it, but the fundamental issue persists: users assume that sending to an address on one chain is equivalent to sending to the same address on another chain. It is not. The address is deterministic, but the state is not. The fix is simple: wallet providers should warn before transactions reach no-code destinations. Yet the study shows that 469 contracts have been deployed to exploit this gap. That is a failure of the user interface layer, not the protocol.
The second vector uses EIP-7702 to make the drain more direct. An attacker uses an exposed key to delegate the account to malicious code that forwards a deposit to the attacker in the same transaction. The detailed analysis identified more than 17,200 delegated addresses and losses of 25.86 ETH plus 33.45 BNB. EIP-7702 was designed to improve account abstraction, but it introduces a new attack surface when private keys are exposed. The delegation mechanism is powerful, but it amplifies the risk of a single key compromise.
In my deconstruction of Curve Finance's 3Pool in 2020, I found that parameterized fee structures created a subtle arbitrage vulnerability. The lesson was that mathematical elegance does not guarantee financial safety. The same applies here: EIP-7702's elegance in enabling account abstraction does not eliminate the underlying risk of exposed keys. The study shows that attackers are actively exploiting this delegation feature. The 17,200 delegated addresses are a ticking time bomb; each one is a potential drain point.
Contrarian: What the Bulls Got Right
The study's precision rate of 99.11% is remarkable. The researchers randomly sampled inferred cases, had two independent judges, and treated addresses derived from public private keys as confirmed. This methodology is rigorous. The bulls might argue that the $574 million figure, while inflated by price timing, represents a real liability. The two active vectors are only 2.7% of the total, but the rest of the addresses are still risky. The researchers disclosed findings to wallet developers and exchanges. Remediation is underway.
However, the contrarian angle is that the headline figure masks the real problem. The $574 million is an aggregate of detected misuse, not a measure of actual losses from the two active vectors. The study itself admits that the full dollar estimate is not directly caused by the two vectors. The bulls are right to focus on the broader dataset, but they are wrong to treat the $574 million as a single loss number. It is a risk estimate, not a confirmed theft.
Furthermore, the study does not provide a complete remediation rate or a current funded-address count for all 65,340 instances. That is a significant gap. Detection without remediation is just noise. The researchers have done the hard work of identifying the addresses, but the responsibility now falls on wallet developers, exchanges, and users. The bulls might say that the ecosystem is already moving to fix these issues, but the evidence is thin. The 17,200 delegated addresses are still out there. The 469 malicious contracts are still deployed.
Takeaway: Accountability and the Cost of Inaction
Ledger integrity precedes market sentiment. The study is a wake-up call, but it is not a crisis. The two active vectors are small in dollar terms, but they are large in structural significance. The real risk is not the $574 million figure; it is the 65,340 addresses that are still vulnerable. Users can reduce the immediate risk by checking both the address and chain against official sources. Developers should keep test accounts and hardcoded keys out of production. Wallet providers can warn before transactions reach no-code or exposed-key destinations.
Audits reveal what code conceals. The study's methodology is sound, but the industry's response will determine whether the $574 million becomes a liability or a lesson. Precision is the only risk mitigation. The numbers are clear: 3,472.23 ETH and 465.24 BNB are directly attributable to active attack vectors. The rest is a fog of potential. The question is not whether the ecosystem can fix these issues. The question is whether it will act before the next wave of attacks scales the vectors beyond the $15.7 million figure.
Stability is a calculated illusion. The current price of ETH and BNB will change. The losses will compound. The 65,340 addresses are a snapshot of a system that prioritizes speed over security. The $574 million is a reminder that the cost of inaction is not a static number. It is a function of time, price, and the speed of the actors who exploit these structural flaws. The study is a tool, not a tombstone. The industry must use it to build, not to mourn.