A Billion-Dollar Blind Spot: What Compliance Teams Miss in DeFi Transaction Monitoring

Written By: Gal Arad Cohen and Roni Sidi

Executive Summary

Onchain activity carries a distinct risk profile for anti-money laundering and sanctions compliance programs. Transactions settle immediately and cannot be reversed, and contamination introduced at a single point of entry can reach market participants with no dealings in the underlying incident, because illicit funds flow through venues many market participants share. That risk compounds as adoption grows, bringing more retail users and institutional capital onchain while regulation evolves. Legal, compliance, and security teams are expected to demonstrate governance suited to that environment, including monitoring and policy enforcement tools that operate in real time.

Most DeFi compliance programs check who holds a position in a market, and a September 2026 incident showed the gap in that check. An attacker moved more than $1 billion of borrowed capital through Aave, one of the most heavily used lending markets in DeFi, to finance a theft from an unrelated protocol, and closed every position before each transaction ended. Any balance check before or after came back clean.

Now picture the same transaction run by a sanctioned entity instead of an exploiter. The onchain picture is identical. The legal exposure is not.

This article walks through the incident, explains why point-in-time snapshots are insufficient, and sets out what an effective compliance program should include to produce a defensible answer when a regulator asks what it saw.

The Compliance Challenge Onchain

The standard controls used by compliance teams today were built for slower-moving financial infrastructure. List-based screening checks addresses against known bad actors, but the addresses behind an active exploit are often only hours old and do not appear on any lists. Point-in-time monitoring checks who holds positions in a market at fixed moments, such as onboarding, at a quarterly review, or after a flagged incident. Sophisticated actors can structure their activity so that no balance is held at any moment a compliance program normally conducts checks.

The gap tends to surface at the least convenient time, when a regulator asks whether the institution had exposure to a designated entity through a given market on a given date. A program built on position checks can say its screening returned no match. It cannot say whether an entity moved value through the market, because it never measured value flow in the first place. There is no record to search and no basis to file, and “we had no visibility into that” is a materially different answer (and an inferior one) from “we looked and found nothing.”

The Balancer-Aave Incident: What Happened in September

On September 6, 2026, an attacker drained roughly $250,000 from Balancer V1, a deprecated version of an early DeFi protocol whose pools still hold funds but can no longer be patched or paused. A bug in the pools’ pricing had been published six days earlier, and copycats were already sweeping the remaining pools. The theft was small by industry standards, but the financing behind it was not. The attacker needed far more capital than the exploit alone would yield. To finance it, the attacker assembled capital through flash loans, loans that are borrowed and fully repaid within a single transaction. The attacker borrowed more than $1 billion in assets through flash loans, routed the funds through three Aave V3 lending markets as collateral, borrowed $788 million more against them, ran the exploit, and returned everything before each transaction ended. A representative transaction on Etherscan shows the full sequence. Aave itself did not suffer a loss and its markets functioned exactly as designed. They simply served as the financing layer for an exploit that occurred entirely within Balancer.

Blockaid flagged the attacker’s address hours before it attacked anything, because its funding traced back to the Tornado Cash mixer, and tracked the stolen funds live as they moved through swaps and back into the mixer. A Blockaid customer monitoring one of the affected Aave markets as part of its compliance program was alerted as the market’s Stolen Funds exposure crossed the threshold it had set as policy. The platform automatically opened an incident and executed response actions, directing the customer to avoid trading against the contaminated market or exit the position. 

The Blind Spot, Explained

The reason position checks saw nothing is structural. Onchain state is only observable between transactions, and the attacker’s positions existed only inside them. In the block before the attack and the block after it, the attacker held effectively nothing in any Aave market. No monitoring system can poll its way around that, because there was never a moment at which the balances carried the evidence. The evidence was in the flows, more than $1 billion supplied and $788 million borrowed, and flows are the one thing a balance check discards.

The check also comes back reassuring rather than empty. Every remaining holder after the attack is a legitimate supplier, so a holder-based review reads the market as clean. A compliance team acting on that output would conclude the markets had no contact with illicit funds that day, and the conclusion would be wrong.

The same pattern applies with far worse consequences to actors other than the exploiters. Suppose the address routing $1 billion through those markets is an OFAC-designated entity rather than an exploiter. The transaction structure is identical, and the residue left behind would be identical. The technique requires no underlying vulnerability, only a flash loan and a smart contract, which in this case cost roughly $4 in gas fees. Value enters as one asset and leaves as another with a reputable protocol sitting in between, and the activity can be accurately described afterward as nothing more than supplying collateral and repaying it. Any screening program based on current positions would find nothing to report. The regulatory consequence changes completely while the onchain picture stays the same, and the existing monitoring tools are simply not enough. Regulators expect organizations to use all reasonable existing tools to meet a compliance program's goals.

OFAC liability is strict, so an institution can be in violation without knowing it. What the institution controls is whether its compliance program was reasonably designed to catch the activity, which OFAC weighs heavily when it decides how to respond. A program that only checks balances could not have caught it.

How Blockaid’s Risk Exposure Closes the Blind Spot

Blockaid built Risk Exposure to see the flows that point-in-time position monitoring misses. It gives compliance teams real-time coverage across the full lifecycle of onchain activity, the deposits they accept, the transactions they sign, and the positions they already hold, and it takes the risk thresholds a team sets as policy and enforces them continuously across every fund flow. 

1. Flows are measured directionally and attributed to entities - Risk Exposure reads a market’s own transfer activity, splits it by direction, supply in and borrowing out, and attributes every movement to the entity behind it, separating direct contact from distant association so a team is never forced to treat the two the same. The attacker’s exit erases nothing, because the record captures what moved rather than what remains, and the measurement is gross throughput rather than a netted figure that collapses back to zero.
2. Verdicts come from the prevention layer - The attacker’s addresses were hours old and on no list anywhere when Blockaid flagged them. Risk Exposure draws its verdicts from the same detection engine that screens live transactions across Blockaid’s network, so an address that shows exploit behavior is flagged from what it does, and the Stolen Funds categorization lands while the activity is still running.
3. Positions remain the second input - A sanctioned entity that still holds a position is live exposure with a counterparty and a balance, a freezing and reporting question. One that passed through and left is a historical contact, still reportable, no longer a balance-sheet item. Risk Exposure weighs both by their dollar magnitude, so a large flow cannot hide behind a quiet holder set and a small historical transfer cannot outweigh a large active position.
4. Thresholds turn measurement into action - Institutions set the exposure level they will tolerate for each market, and a breach opens an incident automatically, with the entity, transaction, market, direction, and amount attached, and with audit-ready breakdowns that support SAR\STR filings and regulator conversations. The same data backs pre-transaction screening, so a proposed interaction is evaluated against both what a market holds and what recently moved through it before anyone signs.

Learn more about Risk Exposure →

Takeaway for Compliance Teams

The Balancer-Aave incident ran end to end through Blockaid’s platform, from the first funding flag to the customer incident, and it makes the takeaway concrete. Sophisticated actors leave flow evidence rather than balances. Every position check came back clean while more than $1 billion moved through markets institutions touch every day, and the risk was assembled at runtime across venues that are each safe on their own.

Detection alone is not a control. Regulators ask what an organization monitored, what it found, how it investigated, and how it remediated a specific event. A defensible program needs flow history that survives a closed position, position data for what remains economically exposed, entity attribution that separates direct contact from distant association, and incident correlation that explains why a number changed and when. Point-in-time monitoring, by design, cannot produce that record. A program that wants a defensible answer needs one that can. 

Request a Demo →

About Gal Arad Cohen

Gal Arad Cohen is a Partner at S. Horowitz & Co., where she leads the firm’s Blockchain, Cryptocurrency and Digital Assets practice. She advises global technology companies, financial institutions, and Web3 businesses on digital asset regulation, stablecoins and payments, custody, DeFi, and AML and financial crime compliance, and her background spans senior legal, compliance, and risk roles in traditional financial services. Her work in FinTech and financial regulation is ranked by Chambers and recognized by The Legal 500.

About Roni Sidi

Roni Sidi is General Counsel at Blockaid, where she oversees the company’s legal and policy affairs. She brings more than a decade of experience supporting high-growth technology companies and publicly traded global enterprises, previously serving as General Counsel at Dazz, a cybersecurity company acquired by Wiz, and as Senior Legal Counsel at SolarEdge, where she led corporate, securities, and M&A matters for the S&P 500 company.