$7.5 Million Out of Thin Air:
Inside the Second Verus Bridge Attack

On July 23, 2026, nearly $7.54 million in assets was withdrawn from the Verus bridge smart contract on Ethereum. This was not a conventional code exploit: the attacker submitted proof of a transfer that appeared to pass verification, prompting the bridge to release millions of dollars even though there was apparently no corresponding collateral on the other side. The attacker immediately converted the proceeds into ETH and obscured the trail through a mixer. Most concerning of all, the same bridge had been exploited in a similar way just two months earlier.

BitOK analysts reconstructed how the attack unfolded. Here is how the attacker managed to create $7.5 million out of thin air.

To understand the attack, start with the basic idea. A cross-chain bridge does not create genuinely unbacked coins. To receive an asset on one network, a user must lock an equivalent asset on another and provide proof that the lock occurred. The Verus bridge moved assets between the Verus blockchain and Ethereum: users locked collateral on one side and received an equivalent asset on the other.

The bridge has an important feature: it is non-custodial. In other words, no company or administrator controls the funds through a set of private keys. Transfers are confirmed by cryptographic proofs verified by the Verus network itself, including miners and stakers. That distinction explains the nature of the attack: there was no custodian to compromise. The bridge's security depended on proof verification, and that is precisely where the attacker struck.

A simple analogy helps explain the exploit. Imagine a container divided into two compartments: Verus on one side and Ethereum on the other. To collect goods from the Ethereum compartment, a customer must deposit goods of equal value in the Verus compartment and present a delivery receipt proving that the deposit was made. The attacker produced a receipt bearing what looked like a valid stamp. The guard checked the stamp but did not verify how many boxes had actually been placed in the other compartment. Believing everything was in order, the guard released the goods.

The key facts:

BitOK's central hypothesis is as follows: before releasing assets, the bridge should have independently verified that the list of recipients and amounts originated from a confirmed transfer on the Verus side. That check appears to have been bypassed. For example, the bridge may have trusted data supplied by the attacker instead of matching it against the underlying transfer. In that scenario, the proof appears valid even though nothing actually backs it. The apparent weakness therefore lies in how the bridge binds a proof to the real transfer it is supposed to represent.

The bridge released a broad mix of assets to the attacker. Note the final row: DAI was not withdrawn from the bridge's reserves; it was minted directly within the malicious transaction.

Here, tBTC is a tokenized form of Bitcoin on Ethereum, EURC is a euro-pegged stablecoin, and scrvUSD and MKR are tokens used in the decentralized finance ecosystem.

The presence of DAI and MKR among the proceeds was not coincidental. Together with ETH and the protocol's native coin, VRSC, they form part of the bridge reserves that fully back its internal asset.

A collection of different tokens is difficult to conceal, so the attacker consolidated them into a single cryptocurrency.

The attacker proceeded as follows:

The hashes of the main attack transactions are provided below for verification in a blockchain explorer:

How the attacker moved the funds. The graph was built using BitOK's Graph tool.

It is important to stress that this was the second attack on the bridge. In May 2026, the same intermediary contract lost approximately $11.6 million in an exploit targeting the same import mechanism. The July incident used a new set of addresses but involved the same contract, the same submitImports entry point, and a similar class of vulnerability.

Investigator's note: the attacker did not target an abandoned legacy version of the code. At the time of the incident, the bridge was running an updated import-processing implementation activated on July 12, more than ten days before the attack. A comparison of the versions shows that the update made few changes to the core import-creation logic. This suggests that the root cause of the May exploit may never have been fully addressed.

The attacker pre-funded gas through a mixer and split operational roles across several wallets. The attacker then submitted a “proof” that the bridge treated as legitimate and withdrew nearly $7.54 million in assorted assets. What followed was a rapid conversion of the entire haul into ETH and 47 Tornado Cash deposits in less than an hour.

Unlike in some other incidents, the funds did not remain parked at a transparent address where they could be monitored. They were sent directly into a mixer, suggesting that laundering had been planned from the outset.

Investigator's note: on-chain data reliably confirms the transaction sequence and movement of funds. The precise root cause of the verification failure, however, remains a hypothesis pending an official post-mortem from the Verus team. There is also no evidence yet that the May and July attacks were carried out by the same individual or group.

The central lesson is that a bridge is only as secure as its verification of external proofs.

When moving assets between networks, users rely not only on the security of both blockchains but also on the correctness of the bridge contract and its verification logic. Before using a bridge, it is worth checking who maintains it, whether it has been audited, and what is known about its incident history.

In this case, the funds went straight into a mixer, leaving virtually no chance of recovery.