Just a few days ago, the Verus-Ethereum bridge suffered a new exploit that ended with the theft of USD 11.58 million. But perhaps the most revealing thing was not only the theft, but the way the crisis was handled publicly. While in the traditional financial system an incident of this magnitude would trigger regulatory hearings, immediate investigations and probably a political and media storm, in DeFi communication about the incident ended up moving to Discord. The project's own official account asked users to follow updates on the hack there.
The scene perfectly sums up one of the great contradictions the DeFi ecosystem carries with it. It tries to present itself as the future global financial infrastructure while continuing to operate, in many cases, with the dynamics of experimental tech communities rather than systemic financial structures.
USD 1 billion so far in 2026
According to data compiled by Observatorio Blockchain, exploits and thefts in DeFi protocols exceed USD 1 billion so far in 2026. Bridges, cross-chain systems and liquid staking protocols have become some of the most vulnerable points in the ecosystem due to their enormous technical complexity and the gigantic amounts of liquidity they concentrate.
The situation began to generate genuine internal concern after the huge exploit linked to KelpDAO and rsETH last May. The attack caused losses of close to USD 293 million and left contaminated positions spread across multiple protocols. For several hours, part of the DeFi ecosystem feared a contagion effect that would end up affecting overall market confidence.
The reaction was as extraordinary as it was revealing. Some of the leading players in decentralized finance, led by Aave, ended up driving DeFi United, a kind of emergency coalition aimed at recapitalizing affected assets and containing reputational damage. So far, the initiative has mobilized close to USD 286 million contributed by more than 116,000 wallets.
Asymmetry between AI and developers
The paradox is hard to ignore. An industry that was born promising to eliminate intermediaries, bailouts and centralized mechanisms has ended up creating collective stabilization funds to prevent systemic crises and restore user trust.
It is in this context that Manuel Aráoz's words take on another dimension. Because the Argentine developer is not simply talking about human error or poor audits. He is arguing that artificial intelligence may have definitively broken DeFi's defensive balance.
Until now, blockchain security worked as a kind of permanent race between developers and attackers. But the emergence of AI agents capable of analyzing millions of lines of code, detecting vulnerable patterns and automating exploit searches at speeds impossible for humans radically changes the rules of the game. Developers need to fully secure a protocol. Attackers only need to find a single flaw. And the more complex DeFi becomes, the harder it is to guarantee that total security.
Offensive artificial intelligence against DeFi
That is probably where the real unease now spreading within the industry lies. It is not only that hacks exist, since they have accompanied the ecosystem for years. What is beginning to worry people is the sense that current decentralized finance architectures may not be prepared to coexist with offensive artificial intelligence capable of searching for vulnerabilities permanently, 24 hours a day, on infrastructures that handle tens of billions of dollars.