MetaMask Validators Exit After Attacker Redirects $967 in Rewards

• MetaMask is removing Ethereum validators from its staking operation after a breach in its infrastructure.
• An on-chain estimate links the exits to about $1.4 billion in staked ETH, a figure MetaMask has not confirmed.
• The attacker redirected a small amount of block rewards and likely never had access to the staked ETH.
• Lido says stETH holders need to take no action, although the exits will cost the pool staking rewards.

MetaMask said on September 30 that it is responding to an ongoing security incident affecting part of its infrastructure and has started “proactively exiting affected validators” from its non-custodial staking business. On-chain analysis ties the exits to about 17,000 Ethereum validators holding roughly 523,000 ETH, close to $1.4 billion at an ETH price near $2,685. The attacker has so far diverted about 0.36 ETH, or $967, in block rewards. MetaMask says it has found no immediate threat to ordinary wallets, but the response now runs through Lido, which uses MetaMask Staking as one of its node operators.

MetaMask has confirmed neither the validator count nor the attack vector. The larger figures come from Ethereum security researcher Kaden, a lead security researcher at Spearbit and Cantina, who traced the affected infrastructure on-chain, and should be treated as estimates until the company publishes its own.

18 of 19 block payments went to a Tornado-funded account

Every validator has a fee recipient, the address that collects priority fees and MEV payments when it proposes a block. Operators set it in the validator client, and for blocks built through MEV-Boost relays the address is declared in messages signed with the validator’s key. Kaden found that 19 MetaMask validators had recently won block rewards, and 18 of those payments went to an account funded through Tornado Cash instead of the correct fee recipient.

The 18 payments only show the validators that happened to propose a block during the attack. Kaden says it is unknown whether the attacker could change fee recipients across the whole set, and MetaMask’s response treats every validator as exposed. Not all of them are out. Three of the exploited validators had still not exited when he published, along with about 821 potentially affected validators in total, for reasons he described as unclear.

Withdrawal keys kept the 523,000 ETH beyond the attacker’s reach

Ethereum splits validator authority in two. The signing key is a hot key that handles attestations and block proposals, while withdrawal credentials decide where the stake and its rewards go when a validator leaves the network, and they can stay offline. MetaMask says it does not manage its clients’ withdrawal keys, so a breach of its operating infrastructure does not hand anyone the stake. Kaden reached the same conclusion from on-chain data, saying the attacker likely never had the ability to withdraw any staked ETH.

For validators with a withdrawal address already registered, a completed exit sends the balance there automatically. That makes exiting the safe option for depositors.

Slashing risk explains a $1.4 billion response to a $967 loss

The signing side is where the danger sits. Kaden warns that, depending on how the attacker obtained signing access, the validators could be slashed on purpose. A key used to sign two conflicting blocks or attestations gets its validator penalized, and Ethereum’s correlation penalty grows with the share of stake slashed in the same window, so a coordinated attack across thousands of validators from one operator would cost each of them far more than an isolated mistake. Taking the validators off duty closes that door while the investigation continues, which is why the three exploited validators still running stand out.

Lido faces up to 45 days of idle stake

Lido confirmed that MetaMask Staking, formerly Consensys Staking, began exiting validators after an investigation into an infrastructure compromise. The protocol estimates the full cycle of exit, withdrawal and re-staking could take up to 45 days, with no rewards and possible downtime penalties along the way. It cites its spread of operators and an ad hoc reserve of more than 6,750 stETH as buffers.

Ethereum’s queues set that timeline. As of October 1, roughly 773,447 ETH is waiting to exit, with a wait of about 13 days, while 1.57 million ETH is queued to activate, a delay of about 27 days. The MetaMask estimate alone would equal about two-thirds of the current exit queue, and much of that backlog may be these validators.

Kiln lost 207 ETH in rewards after a similar exit last year

In September 2025, staking provider Kiln traced an infrastructure breach to a compromised GitHub access token belonging to one of its engineers. No Ethereum validator funds were stolen, yet Kiln exited its validators as a precaution, and Lido’s governance forum later put the cost to Lido at 207.312 ETH in missed rewards, about $557,000 today.

Scaling Kiln’s figure up would be misleading, since queue conditions differ and the 17,000 estimate covers MetaMask’s whole operation rather than just Lido. The longer downtime points to a larger bill all the same.

Six cloud regions offer little cover if credentials are shared

MetaMask’s staking setup splits validators between Teku and Lighthouse, runs Besu and Geth on the execution side, and spreads hosting across AWS and Microsoft Azure in six regions. That design guards against client bugs and cloud outages. It does little when one credential store or deployment pipeline reaches every validator, which is how Kiln was breached. MetaMask has not said whether the same applies here.

What stETH holders and staking clients should watch before October 7

For stETH holders, the near-term effect is a small dip in pooled yield. MetaMask Staking’s share of Lido’s validator set will shrink while the ETH moves through the queues, and Lido’s reallocation process will decide which operators take on that stake. Staking clients are likely to ask harder questions about who can push configuration to validator clients and whether fee recipients are monitored for changes.

The next checkpoints are concrete. The last exits should land by October 7, and whether the three exploited validators and the remaining 821 leave the network will show how far MetaMask believes the access extended. A post-mortem naming the attack vector would settle the slashing question, and if Lido follows the Kiln precedent, its governance forum will publish an exact count of missed rewards.