Crypto bridge explained: how it works and why bridges get hacked
How crypto bridges move value between chains: locking, burning, liquidity and message verification. Learn the risks and how to review a transfer route.

Contents
- What does a crypto bridge connect?
- Three models: lock-and-mint, burn-and-mint and liquidity
- Who verifies the source-chain event?
- Why do bridges get hacked?
- Why speed and fees are incomplete comparisons
- A route-checking example dated November 18, 2026
- Limitations and common mistakes
- Frequently asked questions
A crypto bridge is a system that carries verified messages between blockchains and uses them to make assets available on another network. A coin does not physically move between two ledgers: records, claims and balances change. To understand a crypto bridge, identify what happens to the original asset, who verifies that event and what the recipient actually receives.
What does a crypto bridge connect?
Independent blockchains generally do not read each other’s state directly. A destination contract cannot accept a transfer as proven simply because someone supplies a block-explorer link. It needs a defined mechanism for recognizing an eligible source event and authorizing a corresponding action. Ethereum’s bridge documentation explains this additional infrastructure and its associated risks.
The message need not concern a token transfer. It might communicate a governance result or authorize an operation in another application. A bridge with little money in its own contracts can therefore still carry substantial authority. Assess what its messages are allowed to do, rather than treating the value displayed on its homepage as a complete measure of exposure.
Changing networks is also different from exchanging assets. An aggregator may combine bridging and swapping behind one button, but those steps introduce distinct execution conditions. The distinction between protocol, interface and transaction provider, discussed in our DeFi guide, matters here. A familiar interface can route a transaction through infrastructure operated by several unrelated parties.
Three models: lock-and-mint, burn-and-mint and liquidity
In a lock-and-mint arrangement, an original asset is held as backing while its representation is issued elsewhere. The reverse route typically destroys the representation and releases the backing. The important relationship is redemption: who must return the original asset, and under what conditions? Our wrapped Bitcoin guide explores the difference between an original asset and a claim linked to it.
Circle’s native USDC transfers through CCTP use a different pattern. USDC is burned on the source chain and minted on the destination after the prescribed verification process. That transfer does not require a conventional pool of wrapped tokens. It still depends on the issuer, its attestation process and the supported network configuration. Burn-and-mint does not mean that every trust assumption disappears.
| Model | What happens on the source side? | What arrives? | Main question |
|---|---|---|---|
| Lock and mint | Assets remain in backing | A representation | Can it be redeemed for the original? |
| Burn and mint | Units are destroyed | Newly minted native units | Who verifies the burn and authorizes issuance? |
| Liquidity route | Funds are accepted on one side | Available funds on the other | Who supplies liquidity and completes settlement? |
A hypothetical transfer of 100 units through a liquidity route need not create 100 additional units across the system. A provider can pay from an existing balance. Transfer volume, pool turnover and issuance consequently measure different things. If a swap is added, the recipient’s final amount also depends on execution prices. The table describes architectures, rather than ranking them as universally safe or unsafe.
Who verifies the source-chain event?
Some bridges depend on signatures from a selected group. Relevant questions include the approval threshold, operator independence and procedures for replacing keys. Several servers controlled by one organization do not provide the same separation of authority as genuinely independent participants. A multisignature label without an explanation of the signers leaves important questions unanswered.
Other designs verify proofs using light clients. In the IBC client model, clients track another chain’s consensus states so that delivered messages can be checked against those states. A relayer transports information; transporting a message and having authority to declare it true are separate roles. Security still depends on correct client implementation and the underlying consensus assumptions.
Messaging frameworks can also support more elaborate combinations of verification and execution. Chainlink’s CCIP documentation, for example, distinguishes cross-chain data and token transfers and configurable execution properties. Read the documentation for the actual deployment and current version. An old description of a component is not necessarily an accurate map of every later product using the same brand.
Why do bridges get hacked?
A critical failure occurs when the destination authorizes an action without sufficient evidence of the source event. Possible causes include faulty message verification, compromised signing authority or an unsafe contract change. The failure sits between systems: both blockchains may continue producing valid blocks throughout the incident. Normal block production is therefore not proof that a bridge connecting those networks remains sound.
Consider an educational design requiring three signatures from five signers. If three keys are effectively managed by one team through the same access system, the formal signer count overstates independence. Compromise of that shared system could undermine the apparent separation into five entries. This is a hypothetical example, not a description of a current bridge. It illustrates why organizational authority, backup-key handling and the ability to revoke compromised access need examination alongside the mathematical approval threshold.
Administration and upgrades introduce another layer. An audited contract can interact with a changeable component or an external application receiving its messages. Review scope, update authority and emergency-pause powers therefore need to be examined together. An audit demonstrates that particular work was reviewed under a stated scope; it is not an unconditional promise to reimburse future losses.
Some incidents instead begin at the user interface. A counterfeit site may request a token-spending approval or a signature unrelated to the intended transfer. Funds can then be lost without the official bridge protocol being compromised. Distinguish the application address, wallet permissions and actual transaction before assigning a cause. An ordinary cross-chain transfer does not require entering a seed phrase into a website.
Why speed and fees are incomplete comparisons
A transfer has multiple stages: source inclusion, the required finality, message delivery and destination execution. Completing the first stage does not necessarily make funds available at the destination. If the interface stops updating, determine the stage and message identifier before treating another transfer as a repair. Repeating an operation can create an additional transfer rather than resolve the original one.
A canonical OP Stack withdrawal separates initiation, proof and finalization. The Optimism specification ties completion to the relevant challenge period. A faster third-party route may supply liquidity before that canonical process finishes, but its conditions must be assessed separately. The canonical bridge and an alternative route can connect the same networks while exposing users to different counterparties and completion rules.
Costs are similarly layered. A user may pay for source and destination transactions, message processing, liquidity provision and an additional swap. Compare the expected amount received alongside completion conditions, rather than relying on one prominent advertised fee. A low network charge does not necessarily make a route inexpensive if an unfavorable conversion takes place before the transfer.
A route-checking example dated November 18, 2026
CoinLab’s November 18 snapshot at 2:45 AM ET put total stablecoin supply at approximately $288.10 billion, using DefiLlama. That is a market-wide supply figure, not the reserve of a particular bridge or an amount instantly available for withdrawal. Dated indicators in our market pulse provide liquidity context; they do not replace investigation of the specific route.
Suppose the objective is to receive 100 USDC on another supported network. This is an educational amount, not a live service quote. First decide whether the destination requires native USDC or a specific representation. Then verify the destination network and token address. Two similarly named assets may be accepted differently by an application or a deposit address.
A practical review can follow these steps:
- Find documentation for the actual route and verify its networks, asset and contract addresses.
- Identify who verifies the source event and whether the route uses issuance, backing or liquidity.
- Check the complete expected receipt, charges on both sides and conditions for an unfinished transfer or refund.
- Establish whether destination gas is needed to receive or subsequently use the funds.
- Review wallet permissions and retain the source transaction hash and message identifier where available.
- After execution, verify the balance, token address and ability to perform the intended destination action.
A small test can limit the consequences of an address or route mistake. It does not establish resistance to a future security failure, and fixed costs can make it relatively expensive. Recheck the conditions for later transfers as well: contracts, routing providers or execution settings may have changed since the earlier transaction succeeded.
Limitations and common mistakes
Locked value alone is an incomplete safety measure. A large balance may indicate usage, while also increasing the consequences of failure. A long partner list has similar limits: integrating a service does not necessarily mean every named organization has reviewed every current component. Examine the documented responsibilities rather than assuming that brand association supplies a guarantee.
Another mistake is applying an L2’s properties to every bridge connected to it. The rollup, its canonical withdrawal path and a third-party liquidity service can have different trust assumptions. Our Ethereum rollups guide discusses the separation between execution and settlement. Likewise, a healthy source chain does not automatically guarantee redemption of every representation issued around it.
The term noncustodial also needs context. It may describe a user signing from their own wallet without answering who can upgrade contracts or approve messages. A useful assessment identifies which authorities can change the outcome and what recovery process exists. The appearance of the wallet-connection screen cannot answer those questions by itself.
Frequently asked questions
Does a bridge transfer real coins or a copy? That depends on its model. The result can be a backed representation, native issuance after a burn, or a payment from available liquidity. Check the contract and redemption relationship, not just the ticker.
Why did my source transaction succeed without completing the transfer? It may only have initiated the cross-chain process. Finality, message verification or destination execution can still be pending. Review the whole operation rather than the first transaction alone.
Does a successful small test prove a bridge is safe? It verifies some current route and user settings. It does not establish signer independence, proof-verification quality or the safety of later upgrades. Those require a separate assessment.
Are all bridges between the same networks interchangeable? No. The received asset, administration, verification mechanism, timing and redemption terms can differ. Compare routes against a clearly defined outcome: the exact asset and destination address where usable funds should arrive.
This is not investment advice.



Comments