SKN | Crypto Lending Rebounds 55% as DeFi Confronts New Security Risks

Key Points:

• Crypto lending has recovered more than 55% since July, with total lending value locked reaching roughly $56 billion after a sharp contraction in the second quarter.
• The resurgence comes as interconnected DeFi protocols face broader risks from bridges, oracles, governance systems, custodians and AI-assisted attacks.
• Lending platforms are increasingly focusing on containment, asset-level reviews and operational controls alongside traditional smart-contract audits.

Crypto lending is showing renewed momentum after a difficult second quarter, but the recovery is bringing a familiar question back into focus: whether the sector has done enough to address the interconnected risks that have contributed to major losses in decentralized finance.

Galaxy data cited in the source shows that $11.33 billion left the crypto lending sector during the second quarter. The contraction followed a loss of confidence among users after the Kelp DAO exploit in April, which affected the broader DeFi ecosystem and raised concerns about assets being used as collateral across multiple protocols.

Since the beginning of July, however, total lending value locked has increased by more than 55%, reaching approximately $56 billion.

Interconnected Assets Expand the Attack Surface

The recovery has also increased the amount of capital concentrated across lending protocols, potentially raising the consequences of another major exploit.

The Kelp DAO incident illustrated how an attack on one protocol can spread into other parts of the DeFi ecosystem. Attackers exploited a cross-chain route and created approximately 116,500 unbacked rsETH, worth about $290 million at the time. Some of those tokens were subsequently used as collateral on Aave to borrow other assets.

Aave’s own smart contracts were not breached, but deposits reportedly fell by around $15 billion in the aftermath, while the protocol froze its rsETH and wrsETH markets.

Aave Labs CEO Stani Kulechov said the incident reinforced the need to assess the security of an asset’s entire infrastructure rather than focusing solely on the protocol holding it.

When a lending platform accepts an asset as collateral, its risk exposure can extend to the asset’s bridge, oracle, verification infrastructure and issuer.

Security Must Extend Beyond Smart Contracts

The expanding risk surface means traditional smart-contract audits are no longer sufficient on their own.

Thomas Wu, chief financial officer of Bitcoin-backed lender Ledn, said every wrapper, bridge and oracle between a lender and the underlying collateral represents another potential point of failure. Maple CEO Sid Powell similarly argued that lenders should begin with the possibility that a borrower could fail and determine whether assets can be identified, monitored and recovered quickly.

Spark CEO Sam MacPherson said his platform evaluates governance, operational security, collateral quality, liquidity management and dependencies throughout the broader ecosystem.

Spark began phasing out rsETH from SparkLend in January, before the Kelp exploit, after determining that the asset’s relatively low usage and revenue did not justify the additional risk associated with supporting it.

Aave has adopted a similar approach, according to Kulechov, with assets reviewed quarterly and again following material changes. The protocol has also begun an orderly wind-down of six networks that failed to meet its chain-level standards.

Containment Is as Important as Prevention

Lending platforms also need clear procedures for limiting losses when preventive controls fail.

MacPherson said preventing an exploit is only one part of the security challenge. Protocols also need to demonstrate how losses would be contained if an incident occurs.

Operational security adds another layer. SALT Lending CEO Shawn Owen identified key management, access controls and social engineering as major vulnerabilities that conventional smart-contract audits may not detect.

The experience of the 2022 market collapse also demonstrated the dangers of deploying customer assets elsewhere to generate additional yield. Failures involving Celsius, Voyager and BlockFi showed how lending strategies and counterparty exposure can create risks that are not always apparent to users.

Ledn has sought to limit this exposure by keeping client Bitcoin with qualified custodians rather than lending it out to generate additional returns. Wu argued that reducing the number of transactions involving customer assets can also reduce the number of opportunities for something to go wrong.

AI Offers Both Protection and New Risks

Artificial intelligence is emerging as both a potential security tool and another source of risk for DeFi lending.

Aave has begun using AI-assisted security testing alongside conventional processes. Its mutation testing program deliberately introduced vulnerabilities into V4 contracts, with test suites identifying 271 of 304 injected issues.

A separate review using three AI security tools generated 71 findings across Aave’s V3 and V4 codebases. After human review, 20 were considered valid, while 51 were classified as false positives.

Kulechov said AI provides greater breadth and speed but still requires expert oversight. As AI agents increasingly gain the ability to manage capital onchain, their permissions, inputs and decision-making processes could themselves become targets requiring security controls.

Outlook

The recovery in crypto lending indicates renewed demand for DeFi credit, but the sector is returning with a larger pool of capital and a broader interconnected attack surface. Security is increasingly shifting from a narrow focus on smart-contract code toward a more comprehensive assessment of collateral, bridges, oracles, governance, custody, liquidity and operational controls. As AI becomes more deeply integrated into both security testing and onchain financial activity, lenders will need to manage its benefits alongside the additional risks created by automated systems.