Mentioned in This Article
1. A gap between regulated stablecoins and DeFi lending
Federal rules being written under the GENIUS Act will hold stablecoin issuers to strict reserve, capital and redemption standards, but those protections do not follow the dollars into DeFi lending markets. That is the argument made on November 11 by Mike Hathaway and Joseph Quintilian, the co-founders of the lending protocol Ajna, in a guest column on institutional DeFi risk.
Their central claim is that institutions weighing onchain lending are often asking the wrong question. Rather than looking for the safest-seeming market, the pair say, a lender should ask who has the power to change a loan's terms after the money has gone out. A stablecoin backed one-for-one by reserves offers no defense, in their view, against a token-holder vote that rewrites the rules or a price feed that briefly reports the wrong number.
The piece is an opinion column, and the authors run a protocol built around the alternative they recommend. Their argument is nonetheless grounded in a series of recent incidents across several well-known lending platforms, which they use to show how a single faulty input can spread losses.
2. What is the GENIUS Act?
The GENIUS Act is the United States' federal framework for payment stablecoins, and the authors describe it as crypto's flagship piece of regulation. The Federal Reserve recently proposed two rules to put it into practice. Under them, issuers would have to back their tokens with reserves and meet new capital and risk-management standards.
The proposals also set operating requirements. Issuers would need to honor redemption requests within two business days and report the amount of stablecoins they have outstanding every month. Requirements like these are designed to make sure a token marketed as a dollar can be swapped for a dollar on demand, and that regulators and the public can see how much of it exists.
The law takes effect on January 18, 2027, or 120 days after final rules are issued, whichever comes first. Hathaway and Quintilian describe that as a short runway. They also note that stablecoins are gaining publicly readable guidelines and may find wider use at a national level, including in debt servicing and in extending the reach of the US dollar.
Their concern is not with the issuer rules themselves. It is with what happens to regulated dollars once they leave the issuer and are deposited into lending markets that, in the authors' assessment, still have considerable maturing to do.
3. How DeFi lending markets make decisions
In a typical onchain lending market, users deposit assets that others can borrow against collateral. If the value of that collateral falls too far relative to the loan, the position is liquidated: the collateral is sold off to repay the lender. The trigger for that sale comes from an oracle, a service that feeds external prices into the smart contract.
On top of that, holders of the protocol's governance token typically vote on the parameters that shape each market, such as which assets are accepted, how much can be borrowed against them, and at what point liquidation begins. Those votes can take place after loans have already been made, changing the conditions borrowers and lenders signed up for.
Because liquidations happen automatically, the accuracy of those inputs is critical. A smart contract does not pause to ask whether a price looks reasonable or whether a newly approved setting was a mistake. It acts on whatever it is given, and borrowers whose positions are closed out generally cannot reverse the sale once it has executed. That mechanical quality is what lets these markets run around the clock without intermediaries, and it is also what makes a single error so costly.
The authors acknowledge that each of these pieces exists to protect users. Their objection is to what happens when they are combined. Oracles, governance votes and the specialist risk managers who propose parameter changes together concentrate decision-making power in a relatively small group of people and systems.
4. Curation brought institutions in, and concentration with them
Hathaway and Quintilian trace this concentration to a genuine improvement. In DeFi's early years, every depositor effectively had to act as their own risk officer, judging each market's collateral, oracle and settings alone. That was a serious barrier for institutions used to delegating such work.
Curated vaults and standardized oracle feeds changed that. Institutions could choose from a vetted menu of markets, evaluate a single counterparty in the form of the curator, and discuss risk in terms they already used. The authors credit this model with solving DeFi's onboarding problem.
The trade-off, they argue, is that convenience produces concentration, and concentration is where the risk now sits. When a handful of risk managers and price feeds stand behind billions of dollars in loans, one poorly chosen parameter or a momentary price spike can travel into markets that otherwise have nothing to do with each other, setting off liquidations along the way.
5. Four incidents the authors cite
To support the point, the column walks through a run of failures. The first involved Stream Finance in November 2025, which suffered a loss of $93 million. Analysts subsequently traced roughly $285 million of potential exposure to it across lending venues including Euler, Silo and Morpho, along with others.
In February, a governance vote at Moonwell put in place a price feed that valued Coinbase Wrapped Staked ETH (cbETH) at about $1.12, when the token was actually worth around $2,200. The mispricing left the protocol with roughly $1.78 million in bad debt, meaning loans that could no longer be covered by their collateral.
In March, outdated settings in a risk tool used by Aave priced Lido's wrapped staked ether (wstETH) about 2.85% below where it should have been. That relatively small error was enough to set off around $27 million in liquidations.
In September, two minutes of faulty price data at Vesu liquidated 47 positions worth roughly $3 million in total.
The incidents differ in size and cause: a counterparty failure, a governance decision, a stale configuration and a short burst of bad data. The authors' conclusion is that they share one lesson. In each case, a single bad input was all it took.
6. A familiar problem for Wall Street
Hathaway and Quintilian expect traditional finance to recognize the pattern quickly. In their framing, DeFi is rebuilding a too-interconnected-to-fail structure, but without the capital buffers that the Federal Reserve now plans to demand from stablecoin issuers.
The comparison draws on a lesson that regulators and banks absorbed after past crises: losses spread fastest where many institutions depend on the same few points of failure. In the lending markets the authors describe, those shared points are the oracles that price collateral and the curators and governance processes that set the rules.
That dependence, they argue, is what makes the issuer-level guarantees insufficient on their own. A dollar can be fully reserved and promptly redeemable and still be lent into a market where the terms can shift by vote or where the price that determines liquidations can simply be wrong.
7. Who can change a DeFi loan's terms?
This is the question the authors want every institutional lender, and every DeFi depositor, to answer before committing funds. Their reasoning is rooted in how credit is underwritten. Institutional lenders price risk based on a loan's terms, and if a governance vote can change the collateral rules, liquidation thresholds or interest rates after the money has been lent, the loan cannot be properly underwritten at all.
The same applies to infrastructure the lender does not control. Externally managed components such as oracles can malfunction or go offline. The authors say what lenders need instead are markets whose terms no central authority can alter.
They tie this back to the regulatory picture. Regulated issuers will bring dollars onchain with guarantees attached, including full reserves, capital buffers and two-day redemption. Many of those dollars will end up in lending markets, and that, the authors say, is exactly where the guarantees stop applying.
8. The immutable alternative Ajna proposes
Ajna's design, as the co-founders describe it, makes any change opt-in rather than imposed. The protocol uses no oracle. Lenders instead choose the prices at which they are willing to lend.
Interest rates are set by a fixed formula written into the contract rather than by a vote, and once a lending pool is live, its rules cannot be modified. Anyone can create a market for any asset, with or without a price oracle, and every market runs on the same rules.
The authors are direct about the cost of that approach. A flaw in an immutable system cannot be patched in place. It can only be replaced by a new version, which users must choose to move to themselves. In return, a lender knows that the conditions in force on the day of the loan are the conditions that will apply until it is repaid.
9. What this means for institutions entering DeFi
The column lands as the GENIUS Act's implementation clock is running, with the January 18, 2027, effective date approaching unless final rules arrive sooner and start the 120-day countdown. As compliant stablecoins scale, institutions will face decisions about where to deploy them, and onchain lending is one of the most obvious destinations.
Hathaway and Quintilian's message to those institutions is that regulatory clarity at the issuer level should not be mistaken for safety at the lending level. The questions that matter, in their view, concern who holds the keys to a market's parameters and what happens if a price feed fails.
They close by revisiting what decentralized finance was meant to achieve. The goal, they write, was never to find better gatekeepers. It was to need fewer of them.