Worked example · 2026
Q1–Q2 2026 Sub-Million Exploit Cohort — 36 Incidents Across 13 Chains — Aggregate ~$10.1M
Summary
Timeline
| When | Event | OAK ref |
|---|---|---|
| 2026-01-01 | PRXVT: Sybil attack with CREATE2 addresses on Base; $97K extracted via address-prediction gaming | T9.004 (CREATE2 address manipulation) |
| 2026-01-06 | Fusion by IPOR: EIP-7702 delegation exploit on Arbitrum; attacker leverages delegate-account capabilities; $336K | T9.004 (EIP-7702 delegation abuse — structurally adjacent to smart-account supply-chain surface) |
| 2026-01-20 | ZeroLend: insolvency event on Ethereum; $371K extracted via protocol insolvency mechanism — the protocol became undercollateralised and the shortfall was extracted | T9.004 (insolvency-mechanism exploit — missing solvency-check on withdrawal path) |
| 2026-01-29 | Revert Lend: staked collateral exploit on Base; $50K via collateral-valuation gap | T9.004 (collateral-validation access-control gap) |
| 2026-02-21 | Veil Cash: misconfigured Groth16 verifier on Base; $5K extracted via zero-knowledge proof verification bypass | T9.004 (ZK-verifier misconfiguration — missing constraint in proof verification) |
| 2026-02-26 | Ploutos Money: misconfigured oracle exploit on Ethereum; $390K | T9.001 (oracle misconfiguration) |
| 2026-02-28 | Wise Lending V2: flashloan exploit across Arbitrum and Ethereum; $66K | T9.002 (flash-loan capital acquisition) |
| 2026-03-02 | Curve LlamaLend: donation attack on Ethereum; $240K extracted via vault-share-price manipulation | T9.011 (donation-attack vault-share-price manipulation) |
| 2026-03-07 | Molt EVM: onlySpawnerToken modifier hack on Base; $127K via modifier-bypass access control |
T9.004 (modifier-bypass access-control gap) |
| 2026-03-09 | Gondi V3: PurchaseBundler exploit on Ethereum; $230K | T9.004 (bundler-logic access-control gap) |
| 2026-03-12 | Stake DAO: oracle message spoofing across Arbitrum and Base; $176K | T9.001 (oracle message spoofing), T10.002 (cross-chain message forgery) |
| 2026-03-14 | Goose Finance: share-accounting flaw on BSC; $8,435 via integer-arithmetic rounding error | T9.011 (share-accounting precision loss) |
| 2026-03-17 | dTRINITY dLEND: deposit inflation attack on Ethereum; $257K via vault-share-price donation-and-inflation | T9.011 (deposit-inflation vault manipulation) |
| 2026-03-28 | GoonFi: mispricing arbitrage on Solana; $254K | T9.001 (spot-price mispricing) |
| 2026-03-31 | LML/USDT staking protocol: price manipulation attack on BSC; $950K — the largest incident in this cohort | T9.001 (price manipulation), T9.002 (flash-loan capital) |
| 2026-04-09 | Aethir: access control exploit on BSC; $423K | T9.004 (access-control misconfiguration) |
| 2026-04-12 | SubQuery Network: access control exploit on Base; $60K | T9.004 (access-control misconfiguration) |
| 2026-04-13 | Dango: donate-negative-amounts hack on Dango chain; $410K via integer-underflow donation primitive | T9.011 (negative-amount donation — integer-arithmetic boundary violation), T9.004 (missing input-validation) |
| 2026-04-13 | MONA: BurnAddress accounting exploit on BSC; $60,950 via burn-address tokenomics manipulation | T9.004 (burn-address accounting logic gap) |
| 2026-04-14 | Zerion Wallet: hot wallet compromise via social engineering; $100K | T11.001 (hot wallet social-engineering compromise) |
| 2026-04-20 | Juicebox V3: borrowFrom spoof attack on Ethereum; $52K |
T9.004 (borrow-function caller-authorisation bypass) |
| 2026-04-20 | Thetanuts Finance: first-depositor attack on Ethereum; $50K via vault-initialisation share-price manipulation | T9.011 (first-depositor vault-share manipulation) |
| 2026-04-22 | Kipseli: flawed quoting logic on Base; $80K | T9.004 (quoting-logic access-control gap) |
| 2026-04-26 | Scallop Lend: deprecated V2 rewards contract on Sui; $150K via abandoned-contract residual-value drain | T9.004 (deprecated-contract anti-pattern) |
| 2026-04-27 | Singularity Finance: oracle misconfiguration exploit on Base; $413K | T9.001 (oracle misconfiguration) |
| 2026-04-27 | ZetaChain: GatewayEVM contracts exploit across 6 chains (Ethereum, Base, Polygon, Arbitrum, Avalanche, BSC); $300K via cross-chain gateway access-control gap | T9.004 (gateway access-control gap), T10.002 (cross-chain gateway message validation) |
| 2026-04-28 | JUDAO: flashloan exploit on BSC; $228K | T9.002 (flash-loan-enabled exploit) |
| 2026-04-28 | Quant: access control exploit on Ethereum; $138K | T9.004 (access-control misconfiguration) |
| 2026-04-29 | Syndicate: Commons bridge exploit on Base; $380K | T10.002 (bridge message-verification bypass), T9.004 (bridge access-control gap) |
| 2026-05-01 | Bisq: fake Bisq V1 client hack on Bitcoin; $858K via counterfeit desktop-client distribution serving as signing-oracle compromise | T11.001 (fake-client signing-vendor compromise — structurally analogous to counterfeit hardware/software supply chain) |
| 2026-05-01 | Sharwa.Finance: oracle price manipulation on Arbitrum; $32,850 | T9.001 (oracle price manipulation) |
| 2026-05-04 | SmartCredit: flashloan exploit on Ethereum; $72K | T9.002 (flash-loan-enabled exploit) |
| 2026-05-10 | Renegade: unprotected initializer exploit on Arbitrum; $209K via missing initialisation guard | T9.004 (unprotected initializer — missing initializer modifier) |
| 2026-05-11 | INK Finance: whitelisted-address impersonation on Polygon; $140K via whitelist-spoofing access-control bypass | T9.004 (whitelist-impersonation access-control bypass) |
| 2026-05-12 | Aurellion: uninitialized proxy exploit on Arbitrum; $456K | T9.004 (uninitialized proxy — missing proxy-initialisation guard) |
| 2026-05-12 | SQ Protocol: access control exploit on BSC; $346K | T9.004 (access-control misconfiguration) |
Incident summaries by technique family
T9.004 — Access-Control Misconfiguration (18 incidents, ~$3.5M aggregate)
The dominant exploit class in the sub-million cohort, spanning the full T9.004 sub-pattern taxonomy:
T9.001 — Oracle and Price Manipulation (5 incidents, ~$2.0M aggregate)
T9.002 — Flash-Loan-Enabled Exploit (3 incidents, ~$366K aggregate)
Wise Lending V2 ($66K), JUDAO ($228K), SmartCredit ($72K) — all follow the canonical flash-loan-capitalised atomic extraction pattern at sub-million TVL scale.
T9.011 — Precision-Loss and Donation Attacks (4 incidents, ~$967K aggregate)
T10.002 — Cross-Chain and Bridge (2 incidents, ~$556K aggregate)
Syndicate Commons bridge ($380K), Stake DAO oracle message spoofing ($176K cross-chain component). Both exploit cross-chain message-verification gaps at the bridge or relayer layer.
T11.001 — Key and Client Compromise (2 incidents, ~$958K aggregate)
Discussion
The Q1–Q2 2026 sub-million cohort is the statistical baseline for DeFi exploit frequency: 35 incidents in ~20 weeks, averaging 1.75 per week, with T9.004 (Access-Control Misconfiguration) as the dominant class (51%). The exploit primitives are technically identical to those in the >$1M cohort; the loss magnitude reflects the protocol's TVL, not the exploit's novelty.
Several incidents in this cohort are technically notable despite their smaller dollar figures:
Dango's negative-amount donation ($410K) is the cohort's most technically interesting exploit: the attacker exploited an integer-arithmetic boundary violation to "donate" a negative amount, which underflowed into a value-extracting operation. This is a novel T9.011 sub-variant — the donation primitive is preserved (value transferred directly to the vault), but the sign is inverted via integer underflow, turning a deposit into a withdrawal. The incident suggests that donation-attack mitigations (virtual shares, dead shares, minimum-deposit enforcement) must also validate the sign of the donation amount, not just its magnitude.
Veil Cash's Groth16 verifier misconfiguration ($5K) is the cohort's most technically sophisticated exploit despite its trivial extraction: a missing constraint in the zero-knowledge proof verification logic allowed the attacker to forge a proof that the verifier accepted. The incident is a T9.004 sub-pattern applied to a ZK-verifier — the verifier's constraint set is the access-control surface, and a missing constraint is structurally identical to a missing require statement. The $5K extraction reflects the protocol's early-stage TVL, not the exploit's technical severity — a fully-launched protocol with the same verifier bug would have lost orders of magnitude more.
Bisq's fake-client distribution ($858K) is the cohort's largest T11.001 incident and a reminder that Bitcoin's UTXO model does not eliminate key-compromise risk — it shifts it to the client-software distribution layer. The counterfeit Bisq V1 client is structurally analogous to a counterfeit hardware wallet (Ledger, Trezor) or a fake browser extension (MetaMask) — the attack surface is the software supply chain, not the cryptographic primitive.
Fusion by IPOR's EIP-7702 delegation exploit ($336K) is structurally adjacent to the account-abstraction (T13) surface: EIP-7702 allows EOAs to delegate execution to smart-contract code, effectively creating temporary smart accounts. An attacker who exploits the delegation mechanism can execute code in the context of any EOA that has set a delegation. The incident is a leading indicator of the EIP-7702 exploit surface and is classified under T9.004 pending further characterisation of the EIP-7702-specific technique class.
Public references
- LML/USDT, Bisq, Aurellion, Aethir, Singularity Finance, Dango, Ploutos Money, Syndicate, SQ Protocol, Fusion by IPOR, ZetaChain, dTRINITY, GoonFi, Curve LlamaLend, Gondi V3, JUDAO, Renegade, Stake DAO, Scallop Lend, INK Finance, Quant, Molt EVM, Zerion Wallet, PRXVT, Kipseli, SmartCredit, Wise Lending V2, MONA, SubQuery Network, Juicebox V3, Thetanuts Finance, Revert Lend, Sharwa.Finance, Goose Finance, Veil Cash: DeFiLlama classifications
- Cross-reference: T9.004 at
techniques/T9.004-access-control-misconfiguration.md; T9.001 attechniques/T9.001-oracle-price-manipulation.md; T9.002 attechniques/T9.002-flash-loan-enabled-exploit.md; T9.011 attechniques/T9.011-precision-loss-rounding-attack.md; T10.002 attechniques/T10.002-message-verification-bypass.md; T11.001 attechniques/T11.001-third-party-signing-vendor-compromise.md