Crypto Bridge Hack Explained: Oracles, Admin Keys, Multisigs
Lesson 8 of 10 · Advanced
What is a crypto bridge and how does it work?
Blockchains by default know nothing about each other. Ethereum cannot see what happened on another chain. A bridge is a system that moves assets or messages between two chains. There are three common models:
- Lock and mint: the original token is locked in a contract on the source chain, and a “wrapped” version is minted on the destination. The value of the destination version depends entirely on that lock staying intact.
- Burn and mint: the token is burned at the source and minted at the destination. The token’s issuer controls supply across all chains.
- Liquidity network: each chain holds a pool of the token, and the bridge takes from the source pool and pays out of the destination pool.
In every model one central question applies: who verifies that something was actually locked or burned on the source chain? If that verifier is fooled or its key leaks, an attacker can mint tokens on the destination with no backing, or release the locked assets. That is why bridges have been among the biggest hacking targets: a single contract holds the assets of all users.
The single verifier: the Kelp DAO case
According to CoinDesk and Chainalysis, in April 2026 an attacker compromised the computer of a developer linked to Kelp DAO and pushed a forged cross-chain message through the LayerZero system. The rsETH route relied on just one independent verifier (a 1-of-1 DVN configuration), so forging that one signature was enough. About 292 million dollars was lost and rsETH on roughly 20 chains was left without reliable backing.
The technical lesson: many cross-chain messaging systems let each project configure the number of verifiers itself. “Built on a well-known protocol” does not mean a secure configuration was chosen. The question to ask is: how many independent signatures are required, and are the signers really independent of each other?
Another example in this category, according to TRM Labs, is the Liquid network (Blockstream’s sidechain for Bitcoin) in September 2026: about 319 million dollars of bitcoin left through a bug in validator software, and about 85 percent of it was later returned. The layer that moves an asset between two systems, however reputable, is a point of failure.
Oracles: when the price lies
A smart contract cannot see the market price; an oracle supplies it. Liquidation, stablecoin minting and derivatives settlement all depend on this price. Three common failure patterns:
- Single-pool oracle manipulation: if a protocol reads its price from a shallow DEX pool, an attacker can use a flash loan (a loan taken and repaid within the same transaction) to move that pool’s price for a few seconds, borrow or liquidate at the fake price in that moment, and then undo everything.
- Fixed or hard-coded price: in the Stream Finance case (November 2025), according to Pharos, some vaults valued xUSD at a fixed 1 dollar even when its market value had collapsed.
- The price of one specific venue: according to 21Shares and CoinGecko, during the crash of 10 October 2025 the stablecoin USDe fell to about 0.65 dollars on one centralised exchange while DEX pools stayed near one dollar. A system relying only on that exchange’s price liquidated positions that were fine across the rest of the market.
A good oracle averages several independent sources, resists momentary spikes (for example with a time-weighted average) and sets conservative caps for illiquid assets.
Admin keys, multisigs and social engineering
Most protocols are upgradeable: someone holds a key that can change the code, alter parameters or pause the contract. To lower the risk, that key is usually a multisig, for example 3 signatures out of 5 people. But a multisig is secure only if the signers know exactly what they are signing.
- Bybit (February 2025): according to Picus Security citing the FBI, the Lazarus group compromised Safe{Wallet}’s developer infrastructure. Malicious code swapped the transaction the signers saw on screen for a different one. About 1.5 billion dollars was lost, the largest crypto theft to date. The legitimate signers unknowingly gave a genuine signature.
- Drift on Solana (April 2026): according to Chainalysis, after roughly six months of social engineering, members of the security council were deceived into pre-signing Durable Nonce transactions, which have no expiry and can be executed later. About 285 million dollars left in about 12 minutes. The attack has been attributed to North Korea.
The shared lesson: the attacks targeted people and their tools, not the contract’s mathematics. An “audited code” label does not cover this risk. What reduces it: a timelock before changes execute so users have time to exit, independent signers on separate devices, inspecting the raw transaction rather than trusting the interface, and limits on emergency powers.
Common mistakes
- Assuming a wrapped token on the destination chain carries the same risk as the original; bridge risk is added on top.
- Trusting a bridge or protocol only because it is built on well-known infrastructure, without knowing the verifier configuration.
- Ignoring a protocol’s price source; a single-source oracle or hard-coded price can create both unfair liquidations and bad debt.
- Reading “multisig” as “secure”; the Bybit and Drift cases showed signers can be deceived too.
- Overlooking the absence of a timelock; if an admin can change code instantly, you have no chance to exit.
- Holding a large amount on a bridge or destination chain for a long time without real need.
What do volume and data say here?
You can check most of these risks yourself in an explorer:
- Bridge backing: compare the balance locked in the source-chain contract with the total supply of the wrapped token on the destination. If destination supply exceeds the source lock, backing is incomplete.
- Contract owner: on the contract page, see whether it is a proxy (upgradeable) and whether its owner or admin is a single wallet or a multisig. For a multisig, check the number of signatures required.
- Timelock: do changes pass through a contract with a time delay? Queued transactions can be viewed too.
- Oracle price versus market: compare the price the protocol uses with prices on DEXs and large exchanges. A persistent gap means risk of unfair liquidation or bad debt.
- Sudden outflows: in hacks, a rapid outflow of assets from the bridge contract is the first thing visible on-chain.
📊 See real gold futures volume behind these ideas: free order flow course · today's gold volume map
Frequently asked questions
What is a crypto bridge?
A system that moves assets or messages between two blockchains, usually by locking a token on the source chain and minting an equivalent on the destination. Its security depends on the verifiers who confirm the transfer.
Why do bridges get hacked so often?
Because they hold a large amount of assets in one place, and if the verifiers are fooled or their keys leak, an attacker can mint tokens without backing. The Kelp DAO case in April 2026, with a single verifier, is a clear example.
What is an oracle in DeFi?
A source that delivers prices and other off-chain data to a contract. Liquidations and many calculations depend on it; if it gives a wrong or manipulated price, the protocol makes wrong decisions.
Education only, not a recommendation to buy or sell; we never give signals. Price examples are illustrative. Test any method on a demo account first, with strict risk management.