10 Major DeFi Hacks
Netsujo Inc. ·
DeFi (decentralized finance) protocols have lost billions of dollars to hacks since 2020, and the damage is not limited to security incidents in Web3 generally — DeFi accounts for an outsized share of them. This article catalogs ten major incidents by year and loss amount, and groups the attack methods into four vectors: flash loan attacks, private-key or multisig compromise, contract logic flaws, and oracle price manipulation. A recurring finding is that having passed an external audit is not sufficient protection — several of the ten hacked projects had already been audited, sometimes more than once. The article closes with a phase-by-phase security checklist (contract design, PoC/testing, pre-launch, and production operation) intended for teams designing or running a product that integrates with DeFi.
Key takeaways
- Cumulative DeFi hacking losses have reached billions of dollars since 2020, concentrated in four attack vectors.
- Being audited is not enough on its own — several of the ten hacked projects covered here had already passed one or more external audits.
- The four vectors: flash loan attacks, private-key/multisig compromise, contract logic flaws, and oracle price manipulation, each with its own prevention measures.
- A phase-by-phase checklist (design, PoC/testing, pre-launch, production) reduces the risk of repeating a known failure pattern.
- Two of the ten cases (Euler Finance, Poly Network) ended with the attacker returning some or all of the funds — a notable outcome, not something to plan around.
DeFi hacking: the overall picture
Attacks on DeFi protocols are a particularly large share of Web3 security incidents overall, and losses do not stay contained to a single protocol — they often cascade through composability, the way multiple protocols plug into one another. Attack methods grew more sophisticated between 2020 and 2024: early incidents were mostly simple contract logic flaws, while more recent ones increasingly target the design layer itself — distorted economic incentives, abused governance mechanisms, and cross-chain bridge vulnerabilities.
- Cumulative losses: billions of dollars industry-wide since 2020, varying year to year.
- Where attacks concentrate: contract-logic flaws, key management, oracle price manipulation, and bridge design flaws.
- Being audited is not immunity: several audited projects were hacked anyway — an audit is a minimum baseline, not a guarantee. Using multiple audit firms, continuous review in production, and a bug bounty programme are what actually move the needle.
10 major incidents
| # | Protocol | Year | Loss (approx.) | Attack vector |
|---|---|---|---|---|
| 01 | Ronin Bridge (Axie Infinity) | 2022 | ~$600M | Private-key leak / multisig design flaw |
| 02 | Wormhole Bridge | 2022 | ~$320M | Contract logic flaw (signature verification) |
| 03 | BadgerDAO | 2021 | ~$120M | Front-end tampering / key leak |
| 04 | Euler Finance | 2023 | ~$200M | Logic flaw / flash loan attack |
| 05 | Mango Markets | 2022 | ~$110M | Oracle price manipulation |
| 06 | Poly Network | 2021 | ~$610M | Contract privilege-management flaw |
| 07 | Beanstalk Farms | 2022 | ~$180M | Governance attack / flash loan |
| 08 | Cream Finance | 2021 | ~$130M | Oracle manipulation / flash loan |
| 09 | Nomad Bridge | 2022 | ~$190M | Message-verification initialization flaw |
| 10 | KyberSwap | 2023 | ~$47M | Concentrated-liquidity pool logic flaw |
Ronin: validator private keys were leaked via social engineering (phishing), giving attackers control of 5 of 9 validators — a failure of multisig threshold design and validator decentralization. Wormhole: a flaw in signature-verification logic allowed roughly 120,000 wETH to be minted improperly; Jump Crypto covered the loss. Euler and Poly Network are notable because the attackers later returned some or all of the funds, the latter being treated afterward as a white-hat actor — neither outcome should be assumed in advance.
Four attack-vector categories
| Vector | How it works | Prevention |
|---|---|---|
| Flash loan attacks | Borrow a large, uncollateralized sum and run multiple operations inside a single transaction to temporarily distort collateral ratios, liquidity, or oracle prices. Used in Beanstalk, Cream, and Euler. | Time-weighted average price (TWAP) referencing, conservative collateral ratios, building in flash-loan resistance. |
| Private-key leak / multisig design flaw | Validator or operator private keys obtained via social engineering or infrastructure compromise; loose multisig thresholds or weak key custody widen the damage. Ronin and BadgerDAO are the representative cases. | Hardware security modules (HSM), hardware wallets, geographic key distribution, conservative multisig thresholds, regular key rotation. |
| Contract logic flaws | Gaps in signature verification, privilege management, calculation logic, or edge-case handling. More complex economic designs are more prone to being overlooked. Wormhole, Euler, and KyberSwap are representative. | Parallel review by multiple audit firms, formal verification, maintained test coverage, an ongoing bug bounty programme. |
| Oracle price manipulation | Relying on a single external oracle or liquidity pool for price lets an attacker temporarily move the price and distort collateral valuation. Mango Markets and Cream Finance are typical cases. | Multiple oracle references, decentralized oracles (e.g. Chainlink), TWAP, minimum liquidity floors. |
Security checklist for operators
- Contract design phase — choose an upgrade pattern (proxy or diamond); apply least-privilege role management (onlyOwner / role-based); default to multiple oracles plus TWAP for price referencing; set conservative timelock and multisig thresholds.
- PoC and testing phase — confirm unit, integration, and fuzz-test coverage; run penetration testing against likely attack scenarios; get an initial audit from one external firm; treat mainnet candidacy as requiring both internal review and an audit.
- Pre-launch — get a second audit from a different firm (aim for at least two firms confirming any critical-severity finding); start a bug bounty programme; prepare an incident-response runbook (emergency pause, asset preservation); consider an insurance protocol such as Nexus Mutual.
- Production operation — monitor for anomalous transactions and liquidity anomalies; run a security review on a regular cadence (roughly quarterly is a reasonable target); require a pre-audit and staged deployment for any upgrade; keep the bounty programme and its response process running continuously.
Netsujo's security support
Netsujo supports Web3 products from the concept stage through pre-audit internal review and ongoing review in production, including preparing for external audits and helping interpret audit results. We take on enquiries at the concept stage, before requirements are fixed — including "we want to add DeFi integration but are worried about repeating these failures" and "we want a security review of a product that is already live."
Design review informed by these cases, pre-audit internal checks, and ongoing review in production. Enquiries are welcome before requirements are finalized.
Discuss a security design review