Cross-chain Bridges

02 September 2026

Cross-chain bridges are the infrastructure that lets you move tokens and data from one blockchain to another. Without them, each blockchain is a silo: Ethereum has its assets, Arbitrum has its own, Solana has no connection to either. Bridges are what make the multi-chain world usable.

But bridges are also the most attacked category of smart contract in crypto history. Over $2.5 billion has been lost to bridge exploits since 2021. The same mechanisms that make bridges useful - locked collateral, validator signatures, upgradeable contracts - are what attackers exploit.

This page covers how bridges actually work, why they keep getting hacked, and how to use one without losing your funds. Each section points to a dedicated spoke page that goes deeper.

How bridges actually move tokens

The most persistent misconception about bridges is that they "move" tokens. They don't. No token physically leaves one chain and arrives on another. Instead, bridges use one of three mechanisms to represent value on a destination chain.

Lock-and-mint is the most common model. You send tokens to a bridge contract on the source chain, which locks them. The bridge then mints a representation of those tokens - a wrapped asset - on the destination chain. When you want to return, you burn the wrapped tokens, and the bridge releases the originals. This model is used by Wormhole, the Polygon PoS Bridge, and nearly every L1-to-L2 native bridge.

Burn-and-mint works in reverse: you burn tokens on the source chain, and the bridge mints an equivalent amount on the destination. This is common for native token bridges where the same entity controls the token supply on both chains.

Liquidity pool bridges like Stargate and Across take a different approach. Instead of locking your tokens, they swap them against a pool of liquidity that exists on both chains. You deposit USDC on Ethereum; the bridge sends USDC from a pool on Arbitrum. The pool rebalances over time as flows go in both directions. No wrapping, no minting - just a cross-chain swap.

The page How the Lock and Mint Bridge Model Actually Works walks through exactly what happens to your tokens at each step, including where the risk is.

Why Bridges Get Hacked - and What Protects Them

Every bridge faces the same fundamental problem: how does the destination chain know that a deposit really happened on the source chain? The answer determines the bridge's security model - and its attack surface.

Validator multisigs are the most common verification method. A set of trusted parties runs nodes that observe the source chain, agree on what happened, and sign off on it. When enough signatures are collected, the bridge releases funds on the destination. This is how Wormhole, Multichain, and most third-party bridges work. It is also how over $1 billion was stolen: if an attacker compromises enough validator keys, they can sign fake deposits and drain the bridge.

Light clients let the destination chain verify source chain state directly. A bridge contract on Arbitrum can run a simplified Ethereum client that checks block headers and Merkle proofs. No validators, no trusted parties - just cryptographic verification. This is how native rollup bridges work. It is more secure than multisigs but more expensive to operate.

Optimistic verification assumes deposits are valid unless someone proves otherwise during a challenge period. This is how the native Optimism and Arbitrum bridges work, and it is why they have a 7-day wait: time must pass so anyone can submit a fraud proof if the bridge operator cheats.

Zero-knowledge proofs are the newest approach. A prover generates a cryptographic proof that a deposit happened, and the destination chain verifies that proof in constant time. zkSync's native bridge and certain third-party bridges are moving in this direction.

The page How Bridge Validator Multisigs Get Compromised and How to Stay Safe explains exactly how those attacks happen - key theft, social engineering, governance attacks - and what you can check before using a multisig bridge.

The page Why Optimistic Rollup Bridges Have a 7 Day Wait and Faster Alternatives breaks down why that delay exists, what it protects against, and which faster alternatives actually preserve security versus which ones just skip the waiting game.

The architecture decision that affects every transfer

Not all bridges are built the same way. The two dominant architectures - liquidity pool and message passing - behave very differently in practice.

Liquidity pool bridges keep pools of tokens on each chain. When you bridge USDC from Ethereum to Arbitrum, you deposit into the Ethereum pool, and the bridge pays you from the Arbitrum pool. The pools rebalance over time. The advantage is speed: you get your tokens instantly because the liquidity is already there. The disadvantage is that you are exposed to the pool's health. If the pool on Arbitrum runs low on USDC, your transaction fails or incurs high slippage.

Message-passing bridges (also called arbitrary message bridges) send a data packet - a message - from source to destination. That message can be a simple "mint this many tokens" or it can be a complex contract call. LayerZero, Wormhole, and Axelar are message-passing bridges. They are more flexible than liquidity pools because they can carry any data, not just token transfers. But they depend on relayers and validators to deliver the message, which introduces latency and failure modes.

The page Liquidity Pool Bridge vs Message Passing Bridge Compared walks through the tradeoffs and tells you when one architecture clearly beats the other.

What Can Go Wrong - and How to Fix It

Bridges fail in predictable ways. Some are temporary. Some mean your funds are gone.

Transaction stuck pending is the most common issue. Your source chain transaction confirmed, but nothing arrived on the destination. This usually means the relayer hasn't picked up your message, or the destination gas you provided was too low. The page Bridge Transfer Stuck - What to Do When Your Claim Never Arrives gives step-by-step recovery procedures for each major bridge.

"Rate limit exceeded" means the bridge has hit its maximum outflow for that token or route. Rate limits exist to prevent a single large withdrawal from draining the pool in an attack scenario. The page Bridge Rate Limiting - What It Means When Your Transfer Is Throttled explains how limits work, how long they last, and what to do while you wait.

"Insufficient liquidity" on a liquidity pool bridge means the destination pool doesn't have enough tokens to pay you. You can either wait for the pool to rebalance, use a different route, or try a message-passing bridge instead.

"Destination gas too low" means the bridge couldn't execute your claim because you didn't provide enough gas for the destination chain transaction. This is especially common when bridging to networks like Arbitrum or zkSync where execution costs are higher than expected.

The page Common Cross Chain Bridge Errors and Exactly What Each One Means lists all 18 most common error messages, explains what each one actually means, and tells you which ones are fixable and which mean your funds are lost.

The hidden risks you might not see

Some bridge risks are invisible until it is too late.

Token approvals are the most common hidden risk. When you use a bridge for the first time, your wallet asks you to approve the bridge contract to spend your tokens. Many users approve infinite amounts - "unlimited approval" - so they don't have to approve every time. If that bridge contract gets exploited, the attacker can drain every token you ever approved. The page How to Check and Revoke Bridge Token Approvals Before You Get Drained shows exactly how to find and revoke those approvals.

Upgradeable proxy contracts let bridge developers change the contract logic after deployment. This is useful for fixing bugs, but it means the same key that controls upgrades can be used to steal all locked funds. The page Why Bridge Upgradeable Proxy Contracts Are a Hidden Security Risk explains how to check whether a bridge is upgradeable, who controls the upgrade key, and what happens if that key is compromised.

Fake bridge websites are a phishing epidemic. You search for "Stargate bridge," click the first result, and connect your wallet to a site that looks identical to the real one. The fake site asks you to sign a token approval - and drains your wallet. The page Fake Bridge Websites - How Approval Phishing Scams Drain Your Wallet shows how to verify you are on the real site before you sign anything.

Destination chain gas is a problem every new bridge user hits. You bridge USDC to Arbitrum, and now you have USDC but no ETH to pay gas. You cannot swap because you have no ETH to pay for the swap transaction. The page How to Get Destination Chain Gas Tokens When You Bridge to a New Network lists every method to get that first gas token, from faucets to gas airdrops to centralized exchange direct withdrawals.

Choosing the Right Bridge for Your Transfer

Not every bridge is right for every transfer. The decision depends on what you value: speed, cost, security, or convenience.

Native canonical bridges (like Arbitrum's official bridge or Optimism's official bridge) are the most secure option for moving assets to and from a rollup. They use light client verification or optimistic fraud proofs. But they are slow: Optimism's bridge takes 7 days, and Arbitrum's takes about 8 hours. Third-party fast bridges (like Across or Hop) give you instant transfers but add trust assumptions.

Bridge aggregators (like Li.Fi, Socket, and Bungee) compare routes across multiple bridges and give you the best rate. They save you from checking each bridge manually. But they add another contract interaction and a small aggregator fee. The page Bridge Aggregator vs Using a Protocol Directly - Which Gives Better Rates compares the costs and tradeoffs.

CEX withdrawal vs bridge is a decision every new user faces. You can withdraw USDC from Coinbase directly to Arbitrum, or you can withdraw to Ethereum mainnet and then bridge to Arbitrum. The direct CEX withdrawal is simpler, but you trust the exchange with your funds during settlement. The bridge route is more decentralized but more steps. The page Centralized Exchange Withdrawal vs Bridge for Getting Funds to Layer 2 breaks down the cost and safety tradeoffs.

Wormhole vs LayerZero is the most debated comparison in bridging. Both are message-passing protocols, but their security models differ fundamentally. Wormhole uses a validator multisig with 19 guardians. LayerZero uses an oracle (currently Chainlink) and a relayer, with the assumption that these two parties will not collude. The page Wormhole vs LayerZero Security Model Compared Honestly explains what each model actually protects against and where each has failed historically.

Stargate vs Across for USDC bridging is a practical comparison for stablecoin transfers. Stargate uses pooled liquidity and charges a fee based on pool imbalance. Across uses relayers and optimistic verification. The page Stargate vs Across for USDC Bridging - Which Costs Less and Why compares current fees, slippage, and speed for common routes.

How to assess bridge security before you deposit

You can evaluate a bridge's security without being a smart contract auditor. Here is what to check.

Look at the verification mechanism. Does the bridge use a multisig, a light client, optimistic verification, or ZK proofs? Each has different trust assumptions. Multisig bridges are the most common and the most exploited. Light client bridges are the most secure but rare outside native rollup bridges.

Check whether the contract is upgradeable. If it is, who controls the upgrade key? A single EOA (externally owned account) is a red flag. A multisig with known signers is better. A timelock on upgrades is best.

Look at the bridge's history. Has it been exploited before? How did the team respond? Did users get their funds back? Wormhole was exploited for $325 million and was fully reimbursed by Jump Crypto. Multichain was exploited and never recovered.

Check the liquidity depth. For liquidity pool bridges, look at the pool size on both chains. A small pool means high slippage and higher risk of the pool being drained.

Check whether the wrapped token is canonical or bridge-specific. A canonical wrapped token (like USDC.e on Avalanche) is the official version. A bridge-specific wrapped token (like anyUSDC from the old AnySwap bridge) may have limited liquidity and higher depeg risk. The page Canonical Wrapped Token vs Bridge Wrapped Token - How to Tell the Difference shows exactly how to check.

The page How to Check If a Cross Chain Bridge Is Actually Secure Before Using It gives a complete checklist you can run through in under five minutes before any bridge transaction.

What to do if something goes wrong

Bridges fail. The question is whether you can recover.

If your transaction is stuck, do not panic. Most stuck transactions resolve within a few hours. Check the bridge's explorer (Wormhole Portal, LayerZero Scan, or the bridge's own UI) to see if your transaction was picked up by a relayer. If not, you may be able to retry the claim manually.

If your transaction failed due to insufficient destination gas, you need to get gas tokens on the destination chain. This is the chicken-and-egg problem described earlier. The page How to Get Destination Chain Gas Tokens When You Bridge to a New Network has specific solutions for each major chain.

If the bridge itself was exploited, your options depend on the response. Some teams pause the bridge and reimburse users. Others do not. If you hold a wrapped token from an exploited bridge, that token may depeg instantly. Your best move is to swap it for a canonical version or a stablecoin as quickly as possible - but the liquidity may already be gone.

If you approved infinite tokens to a bridge contract that later gets exploited, your tokens are at risk even if you are not actively using the bridge. Revoke the approval immediately. The page How to Check and Revoke Bridge Token Approvals Before You Get Drained shows you how.

The bottom line on bridges

Bridges are necessary infrastructure for a multi