Web3 Security Trends
Updated 15 Sep 2026About these numbers
01 Sep 2025–31 Aug 2026compared with 01 Sep 2024–31 Aug 2025
Attacks and losses over time
Scroll to see every period
View chart as a table
| Period | Reported loss | Attacks | Known amounts |
|---|---|---|---|
| Sep 2025 | $176.5M | 22 | 22 of 22 |
| Oct 2025 | $51.6M | 15 | 15 of 15 |
| Nov 2025 | $288.4M | 18 | 18 of 18 |
| Dec 2025 | $79.5M | 20 | 20 of 20 |
| Jan 2026 | $405.4M | 30 | 30 of 30 |
| Feb 2026 | $36.8M | 14 | 14 of 14 |
| Mar 2026 | $77.8M | 30 | 28 of 30 |
| Apr 2026 | $951.7M | 39 | 38 of 39 |
| May 2026 | $112.5M | 50 | 50 of 50 |
| Jun 2026 | $93.9M | 41 | 41 of 41 |
| Jul 2026 | $299.4M | 42 | 42 of 42 |
| Aug 2026 | $145.2M | 33 | 29 of 33 |
What changed
Compared with the previous period, reported loss fell by $6.50B while incident count rose by 125.
See incidentsDrift Trade drove 11% of known reported loss in this period.
Open incidentSourceIncorrect Share Accounting was the most frequently recorded attack method, appearing in 14 incidents.
Open incidentSource
Attack methods
-
Incorrect Share Accounting14
-
Access Control13
-
Spot Price Manipulation13
-
Improper Access Control11
-
Phishing7
-
Forged Proof6
-
Oracle Issue6
-
Private Key Compromised6
-
Phishing$372.9M
-
Access Control$352.0M
-
Compromised Admin + Fake Token Price Manipulation / Proxy Upgrade Hijack$295.0M
-
Access Control / Cross-Chain Message Spoofing$293.0M
-
Other / Weak Key Generation$130.0M
-
Composable Stable Pools Exploit / Access Control / Rounding Error$128.0M
-
Oracle Issue / Donation Attack$75.0M
-
Delegatecall Exploit / Access Control / Delegatecall Hijack$48.0M
Affected sectors
200 uncategorized incidents excluded
-
DeFi applications115
-
Other21
-
Bridges12
-
Centralized services4
-
Wallets2
-
DeFi applications$788.7M
-
Other$167.5M
-
Bridges$60.1M
-
Centralized services$43.5M
-
Wallets$8.6M
Affected chains
68 unnamed-chain incidents excluded. Multi-chain counts and losses overlap.
-
Ethereum102
-
BNB Chain61
-
Base35
-
Arbitrum33
-
Solana20
-
Polygon14
-
Bitcoin7
-
Sui6
-
Avalanche6
-
Optimism4
-
Ethereum$849.4M
-
Arbitrum$555.7M
-
Solana$474.1M
-
Base$204.8M
-
Bitcoin$176.5M
-
Polygon$134.2M
-
Optimism$130.2M
-
Sonic$128.1M
-
BNB Chain$96.4M
-
Cronos$75.0M
Incidents behind this view
39 matching records · 2026-04 selected · Clear selection
| Date | Incident and attack method | Chain | Reported loss | Evidence |
|---|---|---|---|---|
| 14 Apr 2026 | CoW SwapDNS Hijack | Ethereum | $1.2M | |
| 14 Apr 2026 | Zerion WalletHot Wallet Compromise via Social Engineering / Phishing | Not recorded | $100K | Source |
| 13 Apr 2026 | DangoDonate Negative Amounts Hack / Missing Input Validation | dango | $1.9M | Source |
| 13 Apr 2026 | MONABurnAddress Accounting Exploit / Incorrect Share Accounting | BNB Chain | $61K | Source |
| 12 Apr 2026 | HyperbridgeFake State Proof / Other / Forged Proof | Arbitrum, Base, BNB Chain, Ethereum | $2.5M | Source |
| 12 Apr 2026 | SubQuery NetworkAcces Control Exploit / Improper Access Control | Base | $134K | Source |
| 09 Apr 2026 | AethirAcces Control Exploit / Other / Improper Access Control | BNB Chain | $423K | Source |
| 09 Apr 2026 | HyperliquidOther | Not recorded | $1.5M | Source |
| 07 Apr 2026 | SquidMulticall Approval ExploitationAccess Control / Improper Access Control | Arbitrum, Avalanche, Base, BNB Chain, Optimism | $517K | Source |
| 05 Apr 2026 | DenariaLP accounting rounding asymmetry and unsafe signed-to-unsigned conversion | Linea | $166K | Source |
| 04 Apr 2026 | BSC TMM/USDTReserve Manipulation Attack / Improper Access Control | BNB Chain | $1.7M | Source |
| 03 Apr 2026 | Silo V2Misconfigured Oracle Exploit / Oracle Misconfiguration | Arbitrum | $392K | Source |
| 01 Apr 2026 | Drift TradeCompromised Admin + Fake Token Price Manipulation / Proxy Upgrade Hijack | Solana | $295.0M | Source |
| 01 Apr 2026 | Drift ProtocolAccess Control | Not recorded | $280.0M | Source |
About these numbers
Incident selection
This dashboard covers reported incidents, not every attack or the relative safety of a chain. Each verified canonical incident is counted once. Unverified, rejected, merged duplicate, pending-review, and future-dated records are excluded. Late reports and corrections can change recent periods.
Loss and outliers
Reported loss sums known positive USD amounts. Missing amounts stay unknown, not zero. Median incident loss uses only incidents with a known amount. “Exclude the largest incident” removes the largest known-loss event from both the selected and comparison periods.
Missing classifications
Sector and chain labels come from the incident record. Uncategorized incidents are excluded from sector charts; incidents without a named chain are excluded from chain charts. A multi-chain incident appears under each named chain, so those counts overlap.
Limitations
Reliable malicious-actor attribution is not consistently available, so this page does not rank actor groups. Timing alone cannot show that AI caused an attack, so there is no AI-involvement metric. Confirmed returned funds are too sparsely documented for a recovery trend: 2 of 354 incidents in this view have comparable reviewed accounting, totaling $1.9M.