DeFi protocols have lost $840 million to hacks and exploits through the first five months of 2026, a 70% year-over-year increase over the same period in 2025. Cross-chain bridges account for a disproportionate share: 14 bridge-specific incidents have drained $340.7 million, representing roughly 4...
"We made a mistake by allowing our DVN to act as a 1/1 DVN for high-value transactions." — Bryan Pellegrino, CEO, LayerZero Labs
DeFi protocols have lost $840 million to hacks and exploits through the first five months of 2026, a 70% year-over-year increase over the same period in 2025. Cross-chain bridges account for a disproportionate share: 14 bridge-specific incidents have drained $340.7 million, representing roughly 40% of total losses despite bridges holding a fraction of aggregate DeFi TVL. Bridge TVL has fallen below $45 billion from approximately $50 billion in May, as users withdraw capital in response to repeated security failures.
Two incidents in April 2026 — the $292 million KelpDAO bridge exploit and the $285 million Drift Protocol drain — accounted for $577 million combined. Both were attributed by TRM Labs and Mandiant to North Korean state-sponsored actors, who now account for 76% of all crypto hack value in 2026, up from 64% in 2025 and 39% in 2024. The pattern is clear: fewer attacks, larger targets, and increasingly sophisticated off-chain social engineering rather than smart contract exploitation.
The structural problem is architectural. Cross-chain message verification remains dependent on trusted intermediaries — validator sets, RPC nodes, threshold signature schemes — that introduce human-exploitable single points of failure. Zero-knowledge proof-based bridges offer a mathematically verifiable alternative but remain early-stage. Until that technology matures, bridges remain the weakest link in multi-chain DeFi infrastructure.
Through the end of May 2026, PeckShield tracked 14 major cross-chain bridge exploits totaling $340.7 million in stolen assets. April was the worst single month in DeFi history: $635 million stolen across 28 exploits, roughly quadrupling the $167 million lost during the entire first quarter. Bridge-specific incidents represented 42% of May's $70 million in total exploit losses.
Cumulative bridge losses since 2022 now exceed $2.8 billion, according to data aggregated across multiple security firms. That figure represents approximately 40% of all value hacked in Web3 over the period.
2026 Major Bridge Exploits (Selected):
| Date | Protocol | Amount | Attack Vector | |------|----------|--------|---------------| | Apr 1 | Drift Protocol | $285M | Social engineering / durable nonces | | Apr 18 | KelpDAO (LayerZero) | $292M | RPC node compromise / 1-of-1 DVN | | Apr 10 | Hyperbridge | $12M | Mint function verification flaw | | May 15 | THORChain | $10.8M | Threshold signature scheme exploit | | May 18 | Verus-Ethereum | $11.58M | Forged Merkle proof / missing validation | | Feb | IoTeX ioTube | $4.4M | Private key compromise | | Feb | CrossCurve | $3M | Validation gaps | | Feb | Hyperbridge | $2.5M | Exploit (details limited) |
Total bridge TVL dropped below $45 billion by late June 2026, down from $50 billion in May. Hyperliquid's bridge TVL fell from $4 billion to $341 million in a single month.
The largest DeFi exploit of 2026 did not involve a single smart contract vulnerability. Every contract functioned exactly as designed. The failure was entirely off-chain.
Timeline of compromise: The breach began on March 6, 2026, six weeks before the theft. A North Korean operative socially engineered a LayerZero Labs developer, harvesting session keys that provided access to LayerZero's RPC cloud environment. The attacker then poisoned internal RPC nodes — patching running memory so that LayerZero's own monitoring tools returned correct responses, while the Decentralized Verifier Network (DVN) received falsified data.
On April 18, the attacker executed: a DDoS against external RPC providers forced the DVN to rely exclusively on the two compromised internal nodes. The poisoned nodes reported that 116,500 rsETH had been burned on the source chain (Unichain). No such burn occurred. The DVN, reading from compromised infrastructure, confirmed the fabricated cross-chain message as valid. The Ethereum-side contract released $292 million in rsETH to an attacker-controlled address.
The single-verifier flaw: KelpDAO's bridge operated with a 1-of-1 DVN configuration — a single verifier with no redundancy. LayerZero's documentation recommended multi-DVN setups, but the 1-of-1 configuration remained a common default across integrations. As LayerZero CEO Bryan Pellegrino later acknowledged: "We made a mistake by allowing our DVN to act as a 1/1 DVN for high-value transactions."
KelpDAO disputed this characterization. The protocol claimed LayerZero had approved the 1-of-1 setup. The dispute remains unresolved.
Attribution: Mandiant, CrowdStrike, and independent researchers attributed the attack to DPRK threat actor TraderTraitor (also tracked as UNC4899). The Arbitrum Security Council, coordinating with law enforcement, froze over 30,000 ETH of downstream attacker funds. KelpDAO successfully paused contracts to block a second $95 million theft attempt.
On April 1, 2026, Solana-based DEX Drift Protocol lost $285 million — over 50% of its TVL — in 12 minutes. The attack was the culmination of a six-month in-person social engineering campaign.
Beginning in fall 2025, DPRK-linked operatives used third-party intermediaries (not North Korean nationals) to build face-to-face relationships with Drift team members. A Telegram group was established. Months of substantive conversations about trading strategies and vault integrations followed.
The technical exploitation leveraged Solana's "durable nonces" feature. Attackers manipulated Drift Security Council members into unknowingly pre-signing transactions that transferred admin control. Once in control, the attackers whitelisted a worthless, artificially priced fake token (CVT) as collateral, deposited 500 million CVT, and withdrew $285 million in USDC, SOL, and ETH.
According to TRM Labs, the drain took approximately 12 minutes — making it the second-largest exploit in Solana's history behind the $326 million Wormhole bridge hack in 2022.
The KelpDAO exploit demonstrated that bridge failures do not stay contained. The attacker used 89,567 of the stolen rsETH as collateral on Aave V3 and V4 markets across Ethereum mainnet and Arbitrum, borrowing 52,834 WETH on Ethereum and 29,782 WETH plus 821 wstETH on Arbitrum.
The result: Aave accumulated an estimated $177–236 million in bad debt from unbacked collateral. Over $5.4 billion in ETH fled the platform. Aave's total TVL dropped by approximately $9 billion — from $26.4 billion to $17.7 billion — in two days.
The incident triggered a coordinated recovery effort. Lido Finance, EtherFi, and Aave founder Stani Kulechov proposed putting forward ETH to cover the shortfall. The episode exposed what researchers at the Bank Policy Institute have described as "DeFi runs" — cascading liquidity withdrawals triggered by a single protocol failure propagating through composable financial infrastructure.
This contagion pattern is not theoretical. When bridges fail, the assets they have issued on destination chains become unbacked. Protocols that accepted those assets as collateral inherit the loss. The interconnections that make DeFi composable also make it fragile.
Cross-chain bridges face a fundamental design challenge: verifying that an event on Chain A actually occurred before releasing funds on Chain B. The verification mechanism determines the bridge's security model.
Trusted validator sets rely on a group of human-operated nodes to attest to cross-chain messages. If a sufficient number of validators are compromised — or if the threshold is set to 1-of-1, as in KelpDAO's case — the bridge can be drained. This category includes multisig bridges and DVN-based systems.
Threshold signature schemes distribute signing authority across multiple parties. THORChain's May 2026 exploit ($10.8 million) demonstrated the risk: a newly churned node operator exploited a vulnerability in the GG20 threshold signature implementation, draining funds from a single vault across Bitcoin, Ethereum, BSC, and Base in a coordinated multi-chain operation.
Smart contract validation gaps remain common. The Verus-Ethereum bridge lost $11.58 million because neither side of the bridge validated a crucial field — the input amount on Verus was not checked against the payout amount on Ethereum. The attacker submitted $0.01 worth of VRSC and extracted $11.58 million in ETH, tBTC, and USDC. The missing validation was in the checkCCEValues function.
Security firm Halborn has noted that insufficient validation of cross-chain message provenance accounts for approximately 40% of all bridge exploits since 2022. The vulnerability class persists because every bridge architecture requires some mechanism to confirm message legitimacy, and each mechanism introduces its own attack surface.
DPRK-linked groups now dominate the crypto exploit landscape. According to TRM Labs, North Korean actors accounted for 76% of all crypto hack value in 2026, driven by just two April incidents (KelpDAO and Drift). The two exploits represented 3% of total incident count but the majority of losses.
The escalation trajectory is consistent year over year:
| Year | DPRK Share of Global Crypto Hack Losses | |------|----------------------------------------| | 2020–2021 | Below 10% | | 2022 | 22% | | 2023 | 37% | | 2024 | 39% | | 2025 | 64% | | 2026 (through April) | 76% |
Cumulative attributed theft exceeds $6 billion since 2017. The operational model has shifted from opportunistic smart contract exploitation to months-long social engineering campaigns targeting protocol developers and governance participants directly.
Following the KelpDAO incident, LayerZero implemented several structural changes:
THORChain's automated solvency detection triggered within minutes of its May exploit, halting signing and trading across multiple chains without human intervention. Node operators then stacked manual pauses and cast formal Mimir governance votes to bring the network to a controlled halt within approximately two hours.
These are incremental improvements to fundamentally trusted architectures. They reduce the probability of specific attack vectors without eliminating the underlying trust assumption.
Zero-knowledge proof-based bridges represent the most structurally different alternative. Rather than relying on validators, multisigs, or threshold schemes, ZK bridges generate a cryptographic proof that an event occurred on the source chain. A lightweight verifier contract on the destination chain checks the proof mathematically. No validators to compromise. No RPC nodes to poison. No social engineering attack surface on the verification layer.
Across Protocol's V4 upgrade (July 2025) added ZK verification. Wormhole has begun integrating ZK proofs for certain transfers. Orbiter uses ZKPs for cross-rollup verification.
The technology faces real constraints. Proof generation remains computationally expensive. Full ZK bridge verification at scale is not yet production-ready for all chain pairs. The prover infrastructure itself introduces new trust assumptions unless it can be sufficiently decentralized.
But the direction is clear. As Ethereum researcher Polynya and others in the rollup research community have argued, the only long-term credible bridge design is one based on validity proofs — where the destination chain cryptographically verifies source chain state without trusting any intermediary.
The 2026 bridge exploit data reveals a structural mismatch between how cross-chain infrastructure is secured and how it is attacked. Protocols invest in smart contract audits; attackers compromise RPC nodes, social-engineer developers, and exploit single-verifier configurations. The attack surface has moved off-chain, but the security model has not fully followed.
The economic consequences extend beyond the exploited protocols. The KelpDAO-to-Aave contagion demonstrated that a bridge failure on one protocol can propagate $9 billion in TVL outflows across interconnected lending platforms. For an industry managing $45 billion in bridge TVL, this systemic risk is not adequately priced.
The remediation steps taken — DVN threshold increases, infrastructure replacement, circuit breakers — are necessary but insufficient. They harden specific configurations against known attack vectors. They do not address the fundamental design problem: trusted intermediaries in cross-chain message verification.
Zero-knowledge bridges present the clearest path to trustless verification, eliminating the human-exploitable layer entirely. Until that technology reaches production maturity, cross-chain bridges remain the most economically significant vulnerability in multi-chain DeFi.