Crypto security teams spent most of the last few years bracing for the next reentrancy bug or oracle manipulation exploit. That threat model is now out of date. Data compiled by CertiK and reported by Forbes shows that in the first half of 2026, compromised private keys and wallet infrastructure overtook smart contract code flaws as the single costliest way attackers drain crypto protocols, according to crypto.news. It is the first time the two categories have flipped since tracking firms began publishing annual loss breakdowns.
The numbers back up the shift. DeFi protocols lost at least $1.3 billion to exploits across the first eight months of 2026, and for the first time on record, stolen or misused private keys accounted for more of that total than broken Solidity or Rust code, per the same crypto.news analysis published September 4, 2026. Separate research from QuillAudits puts the first-half 2026 figure even higher: 82.7% of the $935.3 million lost across 87 DeFi incidents traced back to private key compromise or bridge verification failures, not contract bugs. The pattern holds across nearly every major tracker publishing data this year, from TRM Labs to CoinGecko to CoinDesk’s use of DeFiLlama figures, and it is reshaping how exchanges, custodians, and protocol teams budget their security spend heading into 2027.
What “private key compromise” actually means in 2026
The phrase covers more than a hacker guessing a seed phrase. In the incidents tracked this year, key compromise includes stolen signing credentials, hijacked multisig quorum keys, breached admin wallets tied to protocol governance, compromised validator nodes that feed price or state data into a bridge, and social-engineered access to internal infrastructure that eventually exposes a private key. TRM Labs groups these under a single umbrella it calls Infrastructure Attacks, which spans private keys, seed phrases, wallet infrastructure, privileged access, and front-end surfaces. In its 2026 Crypto Crime Report, TRM attributes $2.2 billion, or 76% of tracked losses, across 45 incidents to that category, averaging roughly $48.5 million per incident.
That distinction matters for engineering teams. A smart contract bug is something a static analyzer or a Slither run can catch before deployment. A compromised RPC node feeding false state into a bridge, or an admin key sitting in a poorly segmented CI pipeline, will not show up in a Foundry test suite. The attack surface has moved from the code itself to the operational layer around it, which is a harder problem because it touches DevOps, HR, and vendor management as much as it touches Solidity.
The two incidents that flipped the numbers
Two April 2026 incidents did most of the damage. On April 18, attackers hit Kelp DAO’s liquid restaking token rsETH through a LayerZero-based bridge, draining about 116,500 rsETH worth roughly $292 million. The entry point was not a flaw in the token contract but compromised internal RPC infrastructure that fed manipulated data into the bridge’s omnichain messaging layer, according to reporting from Galaxy Digital’s research desk. Days earlier, on April 1, Drift Protocol on Solana lost about $285 million after attackers obtained compromised administrative keys tied to protocol governance and used them to manipulate collateral pricing.
Together, those two incidents account for nearly 44% of all first-half 2026 DeFi losses, and both were operational failures rather than code-level bugs, a distinction CertiK’s data specifically calls out. Wallet compromise alone was the costliest category of the half, at more than $444 million across the measured incidents, averaging over $13 million per event, the highest average loss of any attack type CertiK tracked.
How the numbers compare across trackers
No two security firms slice the data identically, which is normal for an industry without a shared incident-classification standard. But every major tracker publishing in 2026 points the same direction: infrastructure and key-related losses now outweigh pure smart contract exploits by dollar value, even when the raw incident count still favors code bugs.
| Source | Period covered | Key-compromise share of losses | Total losses tracked |
|---|---|---|---|
| TRM Labs, 2026 Crypto Crime Report | Full-year (infrastructure attacks) | 76% ($2.2B, 45 incidents) | ~$2.9B |
| QuillAudits | H1 2026 DeFi | 82.7% (keys + bridge verification) | $935.3M |
| Decentralized Masters | H1 2026 DeFi | 74% ($789M “OpSec failures”) | $840M |
| MyCryptoParadise | Q2 2026 DeFi (99 exploits) | ~76% (compromised infrastructure) | Not disclosed |
| CoinDesk / DeFiLlama | Cumulative, multi-year | ~40% (private key compromise) | $16.69B |
| crypto.news / CertiK | Jan-Aug 2026 DeFi | Keys overtake contract bugs for first time on record | $1.3B+ |
The spread between 40% (CoinDesk’s cumulative, multi-year DeFiLlama figure) and 82.7% (QuillAudits’ first-half 2026 snapshot) is explained mostly by timeframe. The cumulative number smooths over a decade of exploits, including the early DeFi summer era when reentrancy and flash-loan attacks dominated. The 2026-only figures capture a market where protocols have gotten better at contract auditing but have not caught up on operational security around the humans and infrastructure holding the keys.
A live example: the Liquid Network sidechain incident
The trend was still playing out as this story published. On September 7, 2026, the Liquid Network Bitcoin sidechain disclosed a $405 million incident, after roughly 4,000 of the 4,200 BTC held in a federation reserve wallet moved out, per Straits Times reporting on the Bloomberg-sourced story. This one, however, did not start with a stolen key. Reporting attributes it to a software bug in Elements, the codebase Liquid runs on, which let attackers mint counterfeit LBTC and redeem it for real BTC through the network’s normal peg-out mechanism using an existing SideSwap authorization key. Liquid paused its bridge while it investigates.
That distinction is useful precisely because it is an exception. Most of 2026’s largest losses trace to operational and key-management failures, not logic bugs in the underlying protocol code. Liquid is a reminder that both categories remain live risks simultaneously, and that a team’s threat model needs to cover code-level accounting bugs and operational key custody at the same time, not one or the other.
Why attackers are shifting tactics
The economics are straightforward from an attacker’s perspective. Smart contract audits have become standard practice for any protocol handling meaningful value, and tooling for catching reentrancy, overflow, and access-control bugs has matured. Firms like OpenZeppelin, Trail of Bits, and CertiK itself now review most major launches before mainnet deployment. That has pushed the cost of finding an exploitable contract bug up and the expected payout down, because audited contracts get patched before attackers can use what they find.
Compromising a private key or an admin’s workstation, by contrast, sidesteps the audit entirely. Attackers do not need to find a flaw in verified code. They need to find one employee who reuses a password, one CI pipeline with an exposed secret, or one multisig signer who approves a malicious transaction because the interface displayed misleading data. Blind-signing, where a hardware wallet signs a transaction without showing the signer what they are actually authorizing, has been flagged repeatedly in 2026 postmortems as a contributing factor, though solid adoption data on how many teams have moved to clear-signing interfaces is not yet published by any tracker.
North Korea’s role in the infrastructure-attack wave
State-linked actors, particularly groups tied to North Korea’s Lazarus network, have been named repeatedly across 2026 incident reports as adopters of the infrastructure-attack playbook. Analysts covering the Kelp DAO exploit preliminarily linked it to Lazarus Group activity, tied to the group’s pattern of targeting single-verifier configurations in cross-chain messaging systems, according to crypto.news‘s September 4 writeup, which frames the trend bluntly: compromised keys, not broken code, now drive the majority of crypto theft, and North Korea is cashing the checks.
That framing tracks with years of United Nations and US Treasury reporting on North Korean state-sponsored crypto theft funding weapons programs, but 2026 marks the first year the specific mechanism shifted so heavily toward operational and key-based intrusions rather than the phishing-and-exchange-hack pattern that defined 2022 through 2024.
Smart contract bugs haven’t disappeared, they’ve been demoted
It would be a mistake to read this shift as smart contract security becoming a solved problem. CoinPaprika’s tracking of 15 exploits recorded in 2026 found smart contract bugs, spanning reentrancy, integer overflow, logic flaws, and input validation failures, still caused the highest number of individual incidents. They just did not produce the biggest dollar losses. Decentralized Masters puts the split at 26% of losses from smart contract vulnerabilities versus 74% ($789 million of $840 million tracked) from what it labels OpSec failures: compromised keys, signer compromises, bridge misconfigurations, and backend breaches.
In other words, contract bugs are more common but cheaper per incident, while key and infrastructure compromises are rarer but far more expensive when they land, because they tend to hit privileged accounts capable of moving large reserves in a single transaction rather than draining a contract’s liquidity pool incrementally. That count-versus-cost gap is consistent with the broader 2026 trend, where August alone saw a record 50 separate crypto hacks even as total dollar losses fell from prior months, and where Q2 2026 set a record for DeFi exploit volume at 99 incidents.
2026’s attack categories side by side
Pulling the categories into one table makes the size mismatch between incident count and dollar impact clearer.
| Attack category | Named 2026 example | Approx. loss | Root cause |
|---|---|---|---|
| Compromised admin/governance keys | Drift Protocol (Solana), April 1, 2026 | ~$285M | Privileged keys used to manipulate collateral pricing |
| Compromised bridge infrastructure | Kelp DAO / LayerZero, April 18, 2026 | ~$292M | Compromised internal RPC nodes feeding false bridge data |
| Sidechain code/accounting bug | Liquid Network, Sept 7, 2026 | ~$405M | Elements bug allowed counterfeit LBTC redemption |
| Price oracle manipulation | Aquifer (Solana AMM), Sept 2026 | ~$2.5M | Manipulated oracle price feed |
| Smart contract logic bug | Notional Finance, 2026 | ~$1.73M | Integer truncation in a 2^128 boundary check |
The pattern across the biggest 2026 line items is consistent: the largest checks written to attackers this year came from operational and infrastructure failures, while contract-level bugs, though still common, tend to cap out at far smaller totals because audited protocols catch the biggest logic errors before mainnet.
What CertiK’s data says about where teams should focus
CertiK’s assessment, reported through Forbes and referenced across multiple trackers this year, identifies private key management and multisignature wallet controls as the areas requiring the most urgent attention going into the second half of 2026. During the second quarter specifically, wallet compromises overtook phishing as the largest single attack method, accounting for $807.5 million in stolen assets according to CertiK’s Q2 assessment cited by crypto.news. That is a notable shift from 2024 and 2025, when phishing campaigns targeting individual users and exchange employees were the more commonly cited entry point.
The practical implication for engineering and security teams is that key management now deserves the same rigor as contract auditing. That means hardware security modules or multi-party computation custody instead of single-signer hot wallets for anything holding meaningful treasury value, mandatory clear-signing interfaces so multisig approvers can see exactly what they are authorizing, and infrastructure segmentation so a single compromised RPC node or CI credential cannot reach production signing keys.
Historical context: how the threat model evolved
DeFi’s earliest major exploits, from The DAO hack in 2016 through the flash-loan attacks that defined 2020 and 2021, were almost entirely contract-logic problems: reentrancy bugs, unchecked external calls, and price oracles that could be manipulated within a single transaction. The industry responded by building an entire auditing economy around catching those bugs before deployment, and it worked well enough that pure logic exploits became harder to pull off at scale on audited, widely used protocols.
Attackers adapted. Bridge exploits became the dominant category through 2022, when incidents like Ronin and Wormhole showed that cross-chain messaging systems introduced new trust assumptions that audits often missed. By 2026, the frontier has moved one layer further out, from the bridge’s code to the operational infrastructure and human processes that control the keys authorizing bridge and protocol actions. Each shift has followed the same logic: attackers go where the defensive investment hasn’t caught up yet.
What this means for exchanges and custodians
CoinGecko’s 2026 State of Crypto Security Report states plainly that for centralized exchanges, the most prevalent point of failure remains the compromise of private keys, not application-layer bugs. That finding lines up with the DeFi-side data and suggests the shift is not specific to decentralized protocols. Exchanges, which custody far larger aggregate balances than most individual DeFi protocols, carry outsized incentive for attackers to target their key infrastructure specifically, whether through insider compromise, supply chain attacks on internal tooling, or social engineering against staff with signing authority.
For custodians and exchanges, the response has generally been to push more of their reserves into cold storage with multi-signature or MPC-based approval workflows, and to add mandatory time delays on large withdrawals. None of that eliminates key-compromise risk, but it raises the cost and reduces the blast radius of any single compromised credential. Individual holders face a scaled-down version of the same problem, which is why self-custody setups built around hardware-backed signing have become the baseline recommendation rather than an advanced option.
Market impact and investor response
The shift toward infrastructure attacks hasn’t dented overall market confidence the way a wave of contract exploits might have in 2021 or 2022, largely because the total 2026 loss figures, while still measured in the billions, represent a smaller share of total value locked in DeFi than in prior cycles. CertiK’s broader 2026 dataset shows total losses actually falling roughly 50% year over year even as the composition shifted toward key compromise, according to crypto.news’s coverage of CertiK’s mid-year assessment. That combination, fewer total incidents but a changed root cause, is prompting institutional allocators and protocol treasuries to ask pointed questions about counterparty operational security rather than just requesting a contract audit report before committing capital.
Predictions for the rest of 2026 and into 2027
- Expect due diligence checklists from institutional investors and treasuries to start requiring documented key-management architecture (MPC, HSM, multisig quorum design) alongside the now-standard contract audit report.
- More protocols will adopt clear-signing wallet interfaces by default rather than as an opt-in feature, following pressure from custody providers and hardware wallet makers responding to blind-signing being repeatedly named as a contributing factor in 2026 postmortems.
- Bridge and cross-chain messaging systems will face continued scrutiny of their verifier and RPC node architecture specifically, since the Kelp DAO incident demonstrated that single-verifier trust models remain exploitable even on well-audited contracts.
- Insurance underwriters covering DeFi protocols are likely to start pricing key-management practices into premiums separately from contract-audit status, given how disproportionately key-related incidents have driven 2026’s largest single losses.
- State-linked actors will keep targeting infrastructure and operational weaknesses over contract bugs, since the return on a successful key compromise (measured in the hundreds of millions) dwarfs what most contract-bug bounties or exploits now yield on audited protocols.
What security teams can do now
None of the 2026 data suggests contract audits should get less attention, incident counts show logic bugs remain the most common exploit type even if they’re no longer the costliest. The practical takeaway is additive: teams need both a mature audit process and an equally mature operational security program covering key custody, signer training, infrastructure segmentation, and transaction-review tooling. A protocol with a spotless audit history and a single hot-wallet admin key is still exposed to the exact failure mode that cost Drift Protocol roughly $285 million in April.
Concretely, that means moving treasury and admin functions off single-signature wallets, adopting MPC or hardware-backed multisig with a meaningful signer threshold, auditing the RPC and validator infrastructure that bridges and oracles depend on (not just the contracts themselves), and requiring clear-signing so approvers can verify exactly what a transaction does before signing it. Firms that have historically treated key management as a checkbox item, a hardware wallet in a safe, are the ones most exposed to the pattern CertiK, TRM Labs, and QuillAudits are all independently describing this year.
Frequently Asked Questions
What is the difference between a private key compromise and a smart contract exploit?
A smart contract exploit takes advantage of a flaw in a protocol’s deployed code, such as a reentrancy bug or an unchecked integer calculation. A private key compromise happens when an attacker gains unauthorized control of the cryptographic key that authorizes transactions, whether through theft, social engineering, compromised infrastructure, or insider access, regardless of whether the underlying contract code is flawless.
How much has crypto lost to private key compromise in 2026?
Figures vary by tracker and timeframe. QuillAudits puts private key and bridge-verification losses at 82.7% of $935.3 million in H1 2026 DeFi losses. TRM Labs attributes $2.2 billion (76% of tracked losses) to infrastructure attacks including keys. CoinDesk, citing DeFiLlama’s cumulative multi-year data, puts private key compromise at roughly 40% of a $16.69 billion all-time total.
Were Kelp DAO and Drift Protocol hacked through code bugs?
No. Both April 2026 incidents, which together account for nearly 44% of first-half 2026 DeFi losses, resulted from operational and infrastructure compromises rather than flaws in deployed contract code. Kelp DAO’s roughly $292 million loss traced to compromised internal RPC infrastructure feeding false data into a LayerZero bridge, while Drift Protocol’s roughly $285 million loss traced to compromised administrative keys used to manipulate collateral pricing.
Is the Liquid Network hack an example of key compromise?
No. The September 2026 Liquid Network incident, in which roughly 4,000 BTC (about $320 million) left the sidechain’s federation reserves, has been attributed to a software bug in the Elements codebase that allowed counterfeit LBTC to be minted and redeemed for real BTC. It is a code-level accounting flaw rather than a stolen or misused private key, which makes it an exception to 2026’s broader pattern.
What is blind-signing and why does it matter here?
Blind-signing occurs when a hardware wallet or signing device approves a transaction without displaying its actual contents to the person signing it. Security researchers covering 2026 incidents have repeatedly flagged blind-signing as a contributing factor in key-compromise-adjacent losses, since a signer can be tricked into approving a malicious transaction they cannot actually read. Clear-signing interfaces, which decode and display the transaction’s real effect before approval, are increasingly recommended as a mitigation.
Does this mean smart contract audits are less important now?
No. Contract bugs still account for the highest number of individual 2026 incidents, according to CoinPaprika’s exploit tracking, even though they no longer produce the largest total dollar losses. Audits remain necessary, but the 2026 data shows they are no longer sufficient on their own, since operational and key-management failures now account for the biggest single payouts.
Is North Korea responsible for most of 2026’s key-compromise losses?
Lazarus Group, widely linked to North Korean state-sponsored operations, has been preliminarily connected to the Kelp DAO exploit and named across multiple 2026 incident reports as an adopter of the infrastructure-attack playbook. Attribution for individual incidents is not always confirmed, but researchers covering the trend describe state-linked actors as significant beneficiaries of the shift toward key and infrastructure compromise.
