An attacker spent $1,808 on March 25 to acquire 40.17 million MFAM governance tokens and, within 11 minutes, rammed a malicious proposal through quorum on Moonwell's Moonriver deployment, placing $1.08 million in user funds at risk on a protocol with $85 million in total value locked. The proposa...
"The temp check demonstrated that a single entity holds enough voting power to pass its own budget proposals over community opposition. That same voting power could cancel any active stream at any time." — Marc Zeller, Founder, Aave Chan Initiative
An attacker spent $1,808 on March 25 to acquire 40.17 million MFAM governance tokens and, within 11 minutes, rammed a malicious proposal through quorum on Moonwell's Moonriver deployment, placing $1.08 million in user funds at risk on a protocol with $85 million in total value locked. The proposal, titled "MIP-R39: Protocol Recovery — Admin Migration," sought to transfer administrative control of seven lending markets, the Comptroller, and the Oracle to an attacker-controlled contract.
The incident is the latest in a pattern of governance exploits targeting DeFi protocols with low-liquidity governance tokens and thin voter participation. In a separate but structurally related development, Aave — the largest DeFi lending protocol with $26 billion in TVL — is losing both its primary governance coordinator (Aave Chan Initiative) and its lead engineering contributor (BGD Labs) over disputes about voting power concentration. Combined, these events point to a systemic fragility in on-chain governance models managing an estimated $28 billion in DAO treasury assets across more than 12,000 active organizations.
On March 25, 2026, an unidentified attacker purchased 40.17 million MFAM tokens from the SolarBeam decentralized exchange on Moonriver for 1,600 MOVR — approximately $1,808 at the time of transaction. The MFAM governance token was trading at roughly $0.000025 per token, a price low enough that the attacker could surpass the proposal quorum threshold with a negligible capital outlay.
The attacker then submitted Proposal #74 (MIP-R39), titled "Protocol Recovery — Admin Migration." Blockchain security firm Blockful, which reviewed the on-chain code, stated: "This proposal is clearly an attack," noting that the proposal contract "already includes the transactions necessary to exploit" the seven lending markets and core infrastructure.
The entire sequence — token acquisition, proposal submission, and quorum achievement — took approximately 11 minutes. The proposal, if executed, would have transferred admin keys for seven lending markets, the Comptroller, and the Oracle to an attacker-controlled address, enabling full drainage of approximately $1.08 million in residual user funds locked in the deprecated Moonriver deployment.
Voting closed on March 27 at 10:28 UTC. As of March 26, 68% of votes cast opposed the proposal, with community mobilization shifting the outcome after the attacker's initial quorum. Moonwell maintains an emergency multisig mechanism — the "Break Glass Guardian" — which can override governance execution regardless of vote outcome. Blockful warned that the attacker "can still have hidden wallets, ready to vote in the last block," recommending activation of the Guardian as a precaution.
The ratio is notable: $1,808 in capital to threaten $1.08 million in assets — a potential 597:1 return. The attack targeted Moonwell's deprecated Moonriver instance, but the protocol's main deployment holds $85 million in TVL across its Moonbeam and Base chains, raising questions about whether similar vectors exist on active deployments.
While Moonwell's attack was external, Aave's governance crisis is internal — and arguably more consequential given its $26 billion in TVL.
On March 3, 2026, the Aave Chan Initiative (ACI), the most active governance participant in the Aave ecosystem, announced it would shut down its eight-person team and wind down operations over four months. ACI founder Marc Zeller cited the passage of Aave Labs' "Aave Will Win" proposal — a $51 million budget request in stablecoins plus 75,000 AAVE tokens — as the breaking point.
According to Zeller, ACI had requested four conditions before supporting the proposal: stricter on-chain milestone tracking, limits on self-voting by addresses linked to the budget recipient, enhanced transparency, and accountability benchmarks. None were adopted. The proposal passed its temperature check with approximately 52% support.
Zeller stated: "When we applied those same standards to the entity requesting the largest budget in DAO history, the system stopped working." ACI claimed it drove 61% of governance actions over the prior three years and helped deploy $101 million in incentives.
The exodus deepened when BGD Labs, Aave's primary engineering contributor for four years, announced its departure effective April 1, 2026. BGD cited "changes in the DAO's organizational dynamics and strategic disagreements over the development of Aave v3 and v4." The firm proposed a two-month, $200,000 security retainer to cover incident response through June 1, pending governance approval. Zeller characterized BGD's exit as "the most significant talent loss in Aave's history."
The departures leave Aave Labs — the entity whose budget request triggered the dispute — as the dominant remaining contributor to a $26 billion protocol, a concentration of power that contradicts the decentralization premise of DAO governance.
Governance attacks are not new. They follow a consistent playbook that protocols have been slow to address:
Beanstalk (April 2022): An attacker flash-loaned over $1 billion from Aave, Uniswap, and SushiSwap to acquire enough governance weight to trigger an emergency proposal execution. The attack drained $182 million in collateral; the attacker netted approximately $80 million. The root cause: Beanstalk's emergencyCommit function did not use a flash-loan-resistant mechanism to verify voting percentages.
Build Finance DAO (February 2022): An attacker accumulated enough BUILD tokens to pass a proposal granting full treasury control and unlimited minting authority. The attacker minted 1.1 million BUILD tokens, drained 160 ETH (approximately $470,000), and laundered the proceeds through Tornado Cash.
Tornado Cash (May 2023): An attacker deployed a proposal with hidden self-destruct logic that, once passed, was redeployed with malicious code granting 1.2 million fraudulent voting tokens. The attacker drained 483,000 TORN tokens (approximately $2.17 million) from governance vaults before eventually relinquishing control.
Compound Finance (July 2024): A whale known as "Humpy" spent three months accumulating COMP tokens, then passed Proposal 289 on the third attempt, allocating 499,000 COMP tokens (approximately $24 million) to a yield protocol controlled by Humpy's "Golden Boys" group. A truce was eventually negotiated, with Humpy accepting a counter-proposal for a DAO-controlled staking product.
Cumulative losses from governance exploits since 2022 exceed $250 million. This figure excludes the Compound incident, which was resolved through negotiation rather than outright theft.
Three structural factors make governance attacks economically rational:
1. Low Voter Participation
Across the active DAO ecosystem, fewer than 5% of eligible voters participate in typical governance proposals, according to multiple academic studies and on-chain data analysis. Participation rates average 8–15% of circulating token supply for major proposals among protocol DAOs. Since early 2024, DAO-wide voter participation has dropped by over 40%, with critical proposals passing with only 20–30% of token holders voting. The top 10% of token holders control 76.2% of all voting power, according to governance analytics platform DeepDAO.
Low turnout reduces the capital required to reach quorum. In Moonwell's case, quorum was achievable for $1,808 precisely because the MFAM token was illiquid and the active voting population was negligible.
2. Governance Token Price Collapse
Many DeFi governance tokens have lost 90–99% of their value from all-time highs, a trend accelerated by the Q1 2026 market downturn and the 46-day streak of "extreme fear" in crypto sentiment indices. Depressed token prices lower the cost of accumulating sufficient voting weight. MFAM at $0.000025 per token meant 40 million tokens — enough for quorum — cost less than a dinner for two.
3. Time-Lock and Quorum Design Flaws
Many early DeFi protocols set fixed quorum thresholds denominated in token counts rather than percentages of participating supply. As token prices decline and participation drops, these thresholds become trivially achievable. Emergency execution functions — like Beanstalk's emergencyCommit — compound the risk by allowing single-block proposal execution without mandatory waiting periods.
Protocols have deployed several countermeasures, each with trade-offs:
Time-locks: Mandatory waiting periods between proposal passage and execution give communities time to mobilize opposition. Standard practice is 24–72 hours. Limitation: sophisticated attackers can accumulate hidden wallets and vote at the last block, as Blockful warned in the Moonwell case.
Emergency Multisigs ("Break Glass Guardians"): Moonwell's mechanism allows a designated multisig to override governance in crisis scenarios. Limitation: this is centralized intervention — the antithesis of the trustless governance model DAOs claim to offer.
Flash-Loan-Resistant Voting: Time-weighted average price (TWAP) oracles and multi-block delay requirements prevent single-transaction vote manipulation. Limitation: does not protect against gradual accumulation attacks like Compound's Humpy incident or Moonwell's open-market purchase.
Vote Escrow (ve-token) Models: Protocols like Curve require tokens to be locked for extended periods to gain voting weight, raising the cost and commitment of governance participation. Limitation: creates its own centralization dynamic, as large holders who lock longest dominate governance.
None of these mechanisms address the fundamental problem: when governance tokens lose economic value and voter participation declines, the cost of attacking governance drops below the value of assets under governance control.
The Moonwell attack and Aave's governance exodus expose the same underlying tension: on-chain governance systems were designed for an environment of broad token distribution, active participation, and economically meaningful governance tokens. That environment does not exist for most protocols in 2026.
With over 12,000 active DAOs managing approximately $28 billion in treasury assets and governance token prices at multi-year lows, the attack surface is expanding. The cost-to-attack ratio documented at Moonwell — $1,808 for a shot at $1.08 million — represents not an anomaly but a predictable consequence of depressed token markets meeting low voter turnout.
The defense mechanisms available — time-locks, emergency multisigs, vote escrow — each introduce their own centralization trade-offs. The multisig that saved Moonwell is, by definition, a centralized override of a decentralized process. If protocols require centralized backstops to survive, the governance model they claim to operate under warrants re-examination.
The economic question for protocol designers is straightforward: at what governance token price does the cost of a governance attack fall below the value of assets under governance control? For a growing number of protocols, that threshold has already been crossed.