# Ankr loses $303,000 by reusing an Optimism deposit

> Five withdrawals in three minutes drained Ankr's bridge on Ethereum using an already-redeemed 0.004 ankrETH deposit. The pause flag was still set to false.
> By José Maldonado · DeFi · 2026-11-10

Ankr's bridge between Optimism and Ethereum lost more than $303,000 across five withdrawals executed between 3:09 and 3:12 a.m. Eastern time on November 9. The attacker did not need a new vulnerability in a contract: they reused a legitimate 0.0044024803 ankrETH deposit made on November 3 and already redeemed that same day.

The [chronicle of the reused deposit on Ankr](https://x.com/blockaid_/status/2119712260184527262) reconstructs the method. Each of the five requests carried modified receipt information: a different recipient, larger amounts and different tokens than those in the original transfer. All of them passed the bridge's checks.

Blockaid alerted to the attack at 08:42 UTC, about thirty minutes after the withdrawals were recorded. Its initial estimate was $292,000; later analysis of the stolen tokens raised the figure above $303,000, and two other analytics firms arrived at slightly different figures because they assigned different prices to each token.

In fact, the public contract was still not paused when the first chronicles were published. Forty minutes after the attack, the switch that halts withdrawals remained set to false and no one had explained why. That detail turns a contained loss into an open question about the service's operation.

## What exactly left the contract

The largest component was a withdrawal of 82.91 ankrETH, valued at $216,370. It was followed by another of 515,206 ankrPOL tokens for $56,372 and a third of 7.98 million ANKR for $30,857. Those three pieces add up to the $303,600 that circulates as the official figure for the incident.

There were two more assets that do not enter the count because Etherscan showed no price for them at the time of the assessment: 5,372.25 aMATICb and 0.8196 aETHb. Their value exists, but it is not quantified, so the real total is a floor and not a ceiling.

The five withdrawals left the Ethereum contract without the holdings it had listed. PeckShield estimated initial losses of $286,000 and noted that the attacker converted most of the loot into 103 ether; SlowMist, with its own analysis, raised the figure to about $304,500.

## The mechanism that failed

A bridge of this kind runs on logic that is simple to explain and hard to secure. The user deposits tokens on the source network, sends the transaction for verification and Ankr's backend issues the authorization needed to release assets on the destination network. The deposit is the proof; the signature is the key. The map of [bridges for moving tokens between blockchains](https://observatorioblockchain.com/blockchain/principales-puentes-para-mover-tokens-entre-blockchains-en-defi/) helps to situate the risk, and the [Tendril CoreVault hack](https://observatorioblockchain.com/en/defi/tendril-corevault-exploit-seguro-defi/) showed the same dependence on a central authorization.

The problem is the separation between the two. A valid signature proves who authorized a transaction, not that the deposit backing it is still redeemable. If the system that validates the receipt trusts the signature and does not recheck the deposit's status, a modified receipt with a legitimate signature passes without anyone noticing.

That is the point where the explanations diverge. [A developer's hypothesis](https://defi-planet.com/2026/11/ankr-ethereum-bridge-faces-ongoing-303k-exploit/) points to a possible compromise of the admin key: the bridge would have validated the five withdrawals with its own key. SlowMist maintains that the origin of the signatures remains under investigation and does not confirm that the signer's keys are compromised.

One detail points in that direction without proving it. The associated admin wallet would have been emptied several hours before the attack, which fits prior access to the signing environment rather than manipulation of the contract itself.

## What Ankr has not said

The company had not published a statement with the cause, the scope or a recovery plan at the time of writing, and did not respond to questions about affected services or possible refunds. There is also no confirmation that the incident reached its RPC infrastructure or users' personal wallets.

The pause flag deserves a paragraph of its own. Forty minutes after the withdrawals, the bridge's history showed no new activity, but the public switch that halts withdrawals remained set to false. Blockaid described the exploit as ongoing in its alert, and none of the later sources clarifies whether Ankr took additional measures.

That silence turns a $300,000 loss into a bigger question about the service's operation. A bridge that is not paused in the first half hour leaves open the possibility of a second round, and the only protection the contract had at that moment was that it was already empty.

## Bridges are always the bridges

The incident is not a story about a badly written contract. It is a story about state verification: the system accepted the same proof five times because no one asked again whether it was still valid. Any design that authorizes withdrawals based on a receipt should revalidate it against the original deposit on every attempt.

For the user, the lesson is the usual one and no less true for being repeated. Bridges concentrate risk because they hold funds locked in a contract that other people administer, and the loss figures of recent years are explained by that concentration, not by the quality of each project's code.

That difference matters when choosing where to leave an asset. An audited contract protects against programming errors; a poorly designed authorization process does not show up in a code audit and only appears when someone tests it with an already-redeemed transaction. The second type of failure is the one that leaves no trace until the balance disappears.

For Ankr, the immediate problem is trust. The token had lost value before the incident and the project carries a history of incidents in its infrastructure; a slow response at the moment the flaw is discovered weighs more than any technical note published a week later.

## Three grey points

All the same, three concrete things are still missing. The first is whether the contract was paused at any point and when. The second is whether the signer's keys were compromised or the flaw was in the receipt validation. The third is whether there will be a refund for users who had assets in the bridge.

Until there is an answer, the prudent reading is that the attacker found a door in the authorization process and not in the chain. That is worse news than it seems: a contract flaw is patched and audited, whereas a poorly designed signing process keeps accepting expired proofs until someone rewrites it.
