Every few months a headline announces that another exchange has lost a fortune, and the obvious question follows: how do crypto exchanges get hacked if blockchain technology is supposed to be secure? The honest answer is uncomfortable for the hype but reassuring for the technology: in almost every major incident, the blockchain itself was never broken. Bitcoin’s and Ethereum’s consensus mechanisms have no confirmed case of an attacker rewriting the ledger or forging transactions at the protocol level. What gets hacked is everything built around the chain — the wallets, bridges, smart contracts, signing screens and people that sit between you and the ledger. This article walks through the five attack paths behind the biggest thefts on record, with verified figures, and explains what they mean if your business holds, accepts or builds on crypto.
What actually gets hacked — and what doesn’t
A public blockchain is, at its core, an append-only ledger maintained by thousands of independent computers. To “hack” Bitcoin you would need to overpower that network’s combined computing power or break the cryptography that signs transactions — neither has happened in a way that stole user funds. When investigators publish post-mortems of major thefts, the failure is always somewhere else: a private key stored on an internet-connected server, a bridge contract with a verification bug, a signing interface that showed employees one transaction while executing another, or an employee tricked by a fake job offer. The ledger faithfully recorded every one of those thefts, because from the chain’s perspective they were valid, properly signed transactions. That distinction matters. It means the risk is not evenly spread across “crypto” — it is concentrated in custody and code, and both of those can be engineered properly.
How do crypto exchanges get hacked: the five attack paths
1. Hot-wallet private key theft
Exchanges keep a small share of funds in hot wallets — wallets whose keys live on internet-connected systems so withdrawals can be processed quickly. Those keys are the single most valuable secret the company owns. In May 2019, attackers used phishing and malware to obtain user API keys and two-factor codes and drained Binance’s Bitcoin hot wallet in a single transaction: 7,000 BTC, worth about $40 million at the time and roughly 2% of the exchange’s BTC holdings. Binance’s cold storage was untouched, and its SAFU insurance fund covered users in full. The lesson is structural: the hot wallet is a deliberate, necessary exposure, so its balance has to be capped at an amount the business can afford to lose, and the keys behind it need hardware-backed protection and strict withdrawal allowlists.
2. Cross-chain bridge exploits
Bridges move assets between blockchains by locking funds on one side and minting equivalents on the other — which makes them enormous, concentrated vaults guarded by complex verification code. In October 2022, an attacker exploited a proof-verification bug in the BSC Token Hub bridge and forged a proof that minted 2 million BNB, worth roughly $570 million. Validators halted the chain before most of it could leave; about $100 million was ultimately unrecovered, and the BNB was newly minted rather than taken from user balances — but the incident showed how one bug in bridge code can create money from nothing. Chainalysis estimated that around $2 billion was stolen from bridges across 13 hacks in 2022 alone, roughly 69% of all crypto stolen that year. If your product depends on a bridge, you are trusting its verification logic with everything that crosses it.
3. Smart-contract bugs
Smart contracts are code that holds money, and code has bugs. In August 2021, an attacker exploited an access-control flaw in Poly Network’s cross-chain contracts, effectively registering themselves as a “keeper” authorised to move funds, and withdrew about $611 million — at the time the largest theft in the sector. In a rare twist, the attacker returned nearly all of it within two weeks. Do not plan on that outcome. The normal outcome is the Wormhole and Nomad pattern: a signature-check or initialisation flaw is found, the contract does exactly what the buggy code tells it to do, and the money is gone in minutes. Contracts that custody value need independent audits, formal verification where feasible, and upgrade mechanisms that cannot be triggered by a single key.
4. Signing and interface compromise
The largest theft in crypto history did not steal any keys at all. On 21 February 2025, Bybit lost approximately 401,000 ETH — about $1.5 billion — from a multisignature cold wallet. According to the forensic reviews and the FBI’s attribution to North Korea’s TraderTraitor (Lazarus Group) actors, the attackers compromised a developer machine at Safe{Wallet}, the third-party signing interface Bybit used, and injected malicious code into its front end. When Bybit’s signers approved what the screen showed as a routine internal transfer, the underlying transaction actually handed control of the wallet to the attackers. Three authorised signers approved it in good faith. Multisig only protects you if each signer can independently verify what they are signing — a compromised display turns the strongest vault design into a rubber stamp. Bybit covered all user balances from its own reserves and emergency loans, but the attack redefined the industry’s threat model: the signing path itself is now a primary target.
5. Phishing, social engineering and insiders
The fifth path needs no code exploit at all. The Ronin bridge — the network behind the game Axie Infinity — was drained of about $625 million in March 2022 (173,600 ETH and 25.5 million USDC) after attackers compromised five of its nine validator keys. Reporting by CoinDesk and the subsequent FBI attribution to the Lazarus Group traced the entry point to social engineering — a fake job offer that reached a senior engineer. The theft was not even noticed for six days. Nine validators sounds decentralised; five keys, several controlled by one organisation, is not. The same human layer explains most smaller incidents: support-desk impersonation, malicious browser extensions, SIM-swap attacks on staff accounts, and insiders abusing withdrawal privileges. Technology sets the ceiling on security, but people decide whether you reach it.
Why exchanges are the weak link: centralised keys on decentralised rails
Step back and the pattern is clear. A blockchain distributes trust across thousands of nodes; an exchange re-concentrates it into a handful of keys, a signing workflow and a team of employees. That concentration is not a mistake — customers want instant withdrawals, fiat on-ramps and someone to call when they lose a password, and all of that requires a central operator holding keys. But it means an exchange’s security is only as strong as its key management, its vendors, its code review and its staff training. The BSC Token Hub incident made the same point from the other direction: the chain could only be saved because a small validator set could coordinate a halt — the centralisation that purists criticise was also the emergency brake. When we build trading and payment systems in our trading bot development and platform work, this is the trade-off we design around explicitly: automate aggressively, but keep the keys, the signing path and the kill switch engineered as if they will be attacked — because the record says they will be.
Red flags: what users and businesses should check
You cannot audit an exchange from the outside, but you can look for the signals that separated the survivors from the casualties in every case above:
- Published cold-storage policy. Platforms that state what share of assets sits offline, and prove reserves on-chain, cap their hot-wallet exposure the way Binance’s 2% cap limited the 2019 loss.
- A real insurance or recovery fund. Binance’s SAFU fund and Bybit’s reserves are why those users lost nothing. No fund and vague promises means you are the insurance.
- Withdrawal controls you can set yourself. Address allowlists, withdrawal delays for new addresses, and hardware-key two-factor authentication stop most account-level drains.
- Bridge dependence. If a platform’s yields or transfers depend on third-party bridges, your risk includes those contracts — the single most-exploited category in the industry.
- Incident communication. Both Binance in 2019 and Bybit in 2025 disclosed within hours and kept withdrawals honest. Silence, withdrawal freezes blamed on “maintenance” and deleted posts are the classic pre-collapse pattern.
For businesses that hold crypto on their own balance sheet or accept it as payment, the equivalent discipline is internal: custody design, signer separation and an incident plan written before you need it. Our blockchain security checklist turns that into a working checklist you can run through with your own team.
From our experience: the failures are boring, and that is the good news
Across nine years and 300+ projects — payment integrations, trading automation, custom web platforms — the security failures we are called in to fix are almost never exotic. They are the same five paths in this article wearing different clothes: a secret committed to a repository, an admin panel protected by a password alone, a third-party widget nobody re-audited, a deployment pipeline one person controls end to end. Blockchain systems raise the stakes because transactions are irreversible and settlement is instant, but they do not change the disciplines: least-privilege keys, independent verification of anything that moves value, audited code, monitored everything, and a rehearsed plan for the day something still goes wrong. The teams that survive incidents are not the ones that were never attacked; they are the ones whose architecture assumed the attack.
Frequently asked questions
Has the Bitcoin or Ethereum blockchain itself ever been hacked?
No confirmed incident has broken Bitcoin’s or Ethereum’s consensus or forged transactions at the protocol level to steal funds. Every major theft on record — Bybit, Ronin, Poly Network, Binance 2019 — exploited wallets, bridges, smart contracts, signing tools or people around the chain, not the chain itself.
What was the biggest crypto exchange hack in history?
The February 2025 Bybit theft: approximately 401,000 ETH, worth about $1.5 billion, taken after attackers compromised the Safe{Wallet} signing interface used by Bybit’s multisig signers. The FBI attributed it to North Korea’s TraderTraitor/Lazarus Group. Bybit covered all user balances.
If an exchange gets hacked, do users always lose their money?
Not always — Binance’s 2019 hot-wallet loss was covered in full by its SAFU insurance fund, and Bybit absorbed its 2025 loss from reserves and loans. But reimbursement is the exchange’s choice and capacity, not a guarantee, which is why reserve funds and proof of reserves matter when choosing a platform.
Why are cross-chain bridges targeted so often?
Bridges concentrate large amounts of locked assets behind complex proof-verification code and, in designs like Ronin’s, a small set of validator keys. Chainalysis estimated about $2 billion was stolen from bridges across 13 hacks in 2022 — around 69% of all crypto stolen that year.
How can a business reduce its crypto custody risk?
Keep the large majority of funds in cold storage, require multisignature approvals with independent transaction verification on a second device, audit every contract that can move funds, cap hot-wallet balances, and rehearse an incident-response plan. Our blockchain security checklist article turns each of these into concrete, assignable steps.
Building a product that custodies, trades or moves digital assets — or reviewing the security of one that already does? Talk to our team about the architecture before the attackers review it for you.