Ethereum Founder Vitalik Corrects Multi-Signature Proposal: The Ideal Security Rule is for Each Signer to Change Their Key After Every Operation
Ethereum founder Vitalik Buterin published a correction to the multi-signature proposal: the ideal security rule is for each signer to change their key after every operation. This is a correction, not the release of a new wallet product.
He clarified the premise in parentheses. He assumes that ECDSA may weaken, but not to the point where cracking can be done instantly, so there is no need to design for "being rushed in the memory pool". The real time to compress is another period: the signature has been exposed, but the old key is still valid. He explicitly stated that signatures should be collected offline whenever possible, minimizing the hours between "signature exposure" and "key invalidation".
The latter part of the statement reverses the risk. Hasty or misconfigured upgrades can easily lead to loss of funds, so one should not rush. He did not provide the address for the rotation contract, did not specify what the threshold should change from and to, nor did he say that existing funds should be moved immediately. The ideal rule and executable steps were separated: the rule is to change keys after each operation, and the step is to not rush.
This correction targets multi-signatures where signatures remain valid for a long time once on-chain. Offline collection allows transactions to be assembled before entering the memory pool, with only one completed operation appearing on-chain, while marking the old key as invalid. If ECDSA weakens in the future, the attacker would obtain a key that can no longer sign a transaction, rather than a key that can still control the treasury. The rush only applies when the cracking is fast enough to catch up with the same memory pool, which he ruled out.
The buyers are multi-signature users who manage large treasuries and are concerned about the gradual weakening of the signature algorithm, while the sellers are signers who have been using the same key for many years. This is a correction to the key lifecycle rules, not a response to any theft that has already occurred. Funds will shift from repeatedly used signature keys to one-time-use keys and offline signature collection processes. The beneficiaries are the treasuries that can immediately invalidate old keys after operations, while the pressured ones are those who rush to upgrade and misconfigure thresholds or sign incorrect addresses.
He did not name any specific theft in this correction, nor did he retract his statement that multi-signatures should not be used. The path to losing money was framed as a configuration error, not as an algorithm that has already been broken.
Source: Public Information
ABAB AI Insight
Vitalik Buterin has been promoting smart contract wallets since around 2014, rather than just using external account private keys. After funds were drained from The DAO in 2016, Ethereum used a hard fork to roll back the ledger, and subsequently large treasuries shifted to multi-signatures. Gnosis Safe later became a common treasury shell, with signers still using their own ECDSA keys. This correction is not about the shell, but about the key remaining valid after being used once.
Money movement is halted in the process. Offline signature collection reduces the hours of signature exposure, and after the operation is completed, each signer changes their key, so even if the old key is later computed, it cannot sign another transaction. Account abstraction has long allowed contract addresses to remain unchanged while signers can be replaced; there are also discussions in Ethereum research threads about making ECDSA a one-time key. Failed user operations must also complete the rotation; otherwise, the failure itself exposes the current signer. He now consolidates the same logic into the ideal rules for multi-signatures while keeping the migration budget in check: configuration errors from hasty upgrades can lead to losses before the algorithm weakens.
This is similar to websites changing passwords from being long-term valid to one-time passwords, and banks turning dynamic tokens into single-use verification codes. Passwords can leak, but once leaked, they cannot be used to log in again. Encrypted treasuries are still stuck in the habit of pre-expansion transformation: multi-signatures like Safe have become popular, but one-time key changes have not become the default option. The significant losses related to Ronin in 2022 and Bybit in 2025 were due to the signature path and interface, not because ECDSA was mathematically broken on that day.
Essentially, it is a transfer of pricing power. The security price of the treasury shifts from "how many people sign at the same time" to "how long this key can still be used". The mechanism is that once a signature is made public, the attack clock starts ticking. If cracking is not completed instantly, offline signature collection combined with immediate key changes can stop the clock. The rules can be written as an ideal state first, but if the migration button is rushed and misconfigured, ownership will be lost before cryptography fails.
ABAB News · Cognitive Laws
Once a key has been used, its exposure starts the timer.
Algorithms will gradually weaken, and hasty upgrades will lead to losses first.
Ideal rules can be ahead of time, but migration buttons cannot be.