Over the past 24 hours, three bridge protocols—Verus, AFX, and BSquared—collectively hemorrhaged $35 million. The attacks share a disturbing pattern: every single flaw had been known for months. Verus was drained twice by the same broken cross-chain import validation. AFX's multi-sig keys were abused, a privilege that should have been locked behind hardware security modules. BSquared saw its upgrade rights hijacked after a suspiciously active insider had been lurking for over a year. In a single day, the DeFi bridge ecosystem—a fragile scaffolding that connects the isolated islands of Layer 1s—proved that its most common design choices are not just insecure, but systemically rotten.
This is not a story about clever exploits or black-hat genius. It is a story about hubris dressed in code, where even the most basic lessons of key management and repeatable vulnerability testing are ignored, then papered over with bounties that feel more like ransom notes than responsible disclosure. I have spent years studying governance structures—first in the ICO era, later in the MakerDAO risk committees—and what I see here is a failure not of technology, but of the human systems that surround it. The real question is not how to stop the next hack, but how to stop ourselves from rewarding the same mistakes.
Core Analysis: The Anatomy of a Systemic Collapse
Verus bridge’s story is the most tragic. In May 2024, attackers exploited a flaw in its cross-chain import logic—a flaw that SlowMist had flagged in their audit. The team responded by offering a 25% bounty, recovering 75% of the funds, and presumably patching the code. But patches on the surface do not heal the architecture underneath. Two months later, on July 24, the same root cause was exploited again, draining another $4.5 million. The attacker then funneled the funds through Tornado Cash, ensuring they could never be recovered. This is what happens when a team treats a security audit as a checkbox rather than a process of transformation.
AFX bridge’s $24 million loss illustrates a different but equally insidious failure: the dangerous convenience of privileged roles. The bridge relies on a 5-of-7 multi-signature validation scheme, a design that centralizes trust in a small set of key holders. Attackers managed to acquire and maliciously use one of those authorization keys—likely through sloppy key storage or a compromised device. The fact that the team was able to pause the bridge immediately after the attack (source: AFX’s official post-mortem) confirms that the system is centrally controlled, contradicting the very decentralization that attracts users in the first place.
BSquared’s attack hits closest to my own experience as a governance architect. The attacker exploited an unpatched escalation in the staking contract’s upgrade permissions. Researcher Specter noted that the privileged role responsible had been active for over a year, raising the terrifying possibility of an insider threat. The attacker extracted 8.59 million B2 tokens—worth $3.86 million—and dumped them into a shallow liquidity pool, cratering the token’s value. The team reacted by pledging to compensate affected users, but when a protocol’s core permissions can be hijacked after a year of dormancy, the compensation feels like closing the stable door after the horse has bolted.
What binds these three incidents together is not just the technical vector, but the governance vacuum. Each team held centralized keys, upgrade rights, or multi-sig seats, yet none implemented the most basic safeguards: mandatory hardware wallets for signers, time-locked contract upgrades, or a formal process for rotating privileged roles. In the MakerDAO governance working group I helped lead during DeFi Summer, we spent months debating the ethics of risk parameters—debates that often felt academic until a whale exploit forced us to confront the human cost of our models. These bridge teams never had that reckoning, and their users are paying the price.
Contrarian: Bounties Are a Symptom, Not a Cause
The media narrative around this wave of attacks has focused on whether high bounties (30% for AFX, 25% for Verus) incentivize more hacks. Security researcher Taylor Monahan expressed a common frustration: “Are bounties inviting more hacks?” But I believe this framing misses the deeper rot. A bounty is a reactive tool, a bandage applied after the hemorrhage. The real problem is that the underlying code was never structurally repaired. Verus was vulnerable to the same attack twice—not because the bounty was too high, but because the team never fundamentally rethought their cross-chain validation logic. They paid to recover funds, but they did not pay to recover trust.
In my own work curating small, values-driven DAOs like The Ethereal Archive, I learned that authenticity requires more than a quick fix. When you curate digital artifacts as historical documents, you treat every flaw as a clue to a deeper cultural pattern. The pattern here is clear: teams prioritize speed to market over resilience, and then rely on bounties as an insurance policy against their own negligence. The answer is not to abolish bounties—they are a lifeline for white-hat communities—but to demand that bounties come with a binding commitment to architectural audits and timing delays. If a protocol is exploited twice with the same vector, its bounty should automatically convert to a public disclosure and freeze on new deposits.
Takeaway: Build for Fragility, Not Avoidance
We cannot engineer a perfectly secure system; blockchains are living organisms that evolve through their failures. But we can design governance structures that anticipate fragility rather than denying it. Every upgrade permission should have a sunset clause. Every multi-sig key should be held by independent entities with proven security hygiene. And every bridge that suffers a re-exploitation should be forced into a period of “governance quarantine” until its architecture is independently reviewed by a sworn adversarial testing body.
I have been through the desert of the bear market, watching many promising projects slow-bleed into obsolescence. The ones that survived were those that treated vulnerability not as a secret shame, but as a doorway to deeper alignment. The bridge teams that rebuild after these losses have a chance to show that they have learned—not by offering bigger bounties, but by rewriting their own rules of trust. If they fail, they will join the long list of derivative clones that the soul of this industry must eventually leave behind.