Morning Edition · Saturday, October 31, 2026Published at 1:43 AM EDT · New York
Cardano founder Charles Hoskinson called the sweeper code "seriously dangerous." On the same day, Ethereum researchers posted drafts for short-lived, purpose-limited authorization and for wallet simulations that third parties can check independently.

On Ethereum, the Binance attacker used Ethereum Improvement Proposal (EIP) 7702. It shipped in a 2025 network upgrade to give ordinary wallets smart-contract features. According to CoinDesk, the attacker had obtained 50 signatures from Binance's own wallet. EIP-7702 lets a wallet sign an authorization that points its account to outside contract code. Once the authorization is set, calls to the wallet run that code with the wallet's balances. The attacker pointed a Binance hot wallet to a sweeper contract, which moves many tokens in a few transactions. CoinDesk puts the Ethereum drain at about $2.5 billion, and CryptoTicker reports $2.8 billion. On the other nine chains, the attacker used ordinary signed transfers.
Charles Hoskinson, the founder of Cardano, shared an analysis on X that blames EIP-7702 for the speed of the drain. "Careful, there is some seriously dangerous code," he wrote. Hoskinson leads a competing network, so his criticism serves Cardano's interests as well as the debate. The technical details support a narrower explanation. Every delegation still needed a valid signature from Binance's wallet, and Binance says its keys did not leak. The attacker therefore had signing power through the compromised service. EIP-7702 made the sweep faster and broader, but it did not supply the access. Other chains were drained without any comparable feature.
Three posts on Ethereum Research the same day addressed how wallets grant authority, although none of them cites the hack. One draft proposes scoped capability tokens. These are short-lived, revocable authorizations, signed as EIP-712 typed data or compatible with ERC-1271, that allow one stated action only, such as one transfer to one address. The author says they remove the "ambient authority" that broad signatures carry. A second draft defines a standard, content-addressed simulation result. A hardware device, a second machine or an outside verifier could reproduce it from the same payload and state snapshot. Simulation would then stop depending on a trusted website. A third post sketches post-quantum authorization without signatures, using recursive proofs. It argues that the quantum transition affects how authority is established, not only how signatures are produced.
Part of a tracked trend
Wallet Delegation Becomes an Attack Surface
As wallets gain the power to delegate execution to contract code, each compromise of a signing system will drain more per incident, so authorization design (scope, expiry and independent verification of what is signed) becomes a main front of wallet and custody competition.
Start a discussion in Townsquare.
More from this edition
Binance gains if the delegation feature takes the blame instead of its internal access controls. Rival networks such as Cardano gain a talking point against Ethereum's account-abstraction roadmap when Charles Hoskinson, Cardano's founder, warns of "seriously dangerous code". Ethereum developers gain if the loss is framed as an operations failure.
CoinDesk and Threat Landscape support the narrow reading: Ethereum Improvement Proposal (EIP) 7702 changed what a signed account could run, not who could authorize it. The article understates one harm specific to the feature, though. The attacker submitted 50 authorizations signed on consecutive nonces, and the delegated sweeper kept collecting new deposits for about 3 hours and 50 minutes. Ordinary signed transfers on the other chains could not do that.
An open-source-intelligence read of how likely this story is true with its real nuance, not a judgment of any outlet. It assesses the claim, weighing independent and adversarial reporting. How we label confidence.
What this means
The exposure sits with exchanges, custodians and treasury operators that let hot wallets sign delegation authorizations. Ethereum's account-abstraction roadmap also incurs a reputational cost that rival chains will use against it. Delegation is a sound tool on its own. The new fact is that one compromised signing service can now authorize code that empties every token in a wallet, where before it could only move assets one transfer at a time. Calls to restrict EIP-7702 in the protocol misread the cause. The likelier response comes from operators: within weeks, custody platforms and exchanges add policy rules that refuse delegation authorizations from hot wallets or limit them to approved contracts. Proposals like scoped capability tokens would then move toward wallet standards over months. That depends on Binance's final report confirming that signing access, not a flaw in the protocol, was the entry point. Ethereum core developers formally proposing to limit EIP-7702 in the next hard fork would prove this read wrong.
What to watch
Observations to monitor, not financial advice.
Synthesized from: CoinDesk · crypto.news · Ethereum Research · Ethereum Research · Ethereum Research · CryptoTicker
Comments
0No comments yet.