Dragonfly Managing Partner Haseeb Qureshi Says Cybersecurity Faces a Wave of Attacks Similar to COVID-19
Haseeb Qureshi, Managing Partner at Dragonfly, stated that global cybersecurity is facing a wave of hacker attacks similar to the COVID-19 pandemic, with the crypto industry being the first to experience this trend.
He believes that on-chain attacks are the easiest to monetize directly, and security incidents in crypto protocols can be seen as an early warning for overall internet security risks. Currently, the number of hacker attacks in the crypto industry has sharply increased, but the annualized total amount of stolen funds is roughly equivalent to that of 2025, indicating a decrease in the average loss per attack.
Haseeb pointed out that attackers are focusing on small DeFi protocols with weak security measures and insufficient maintenance, while large protocols have begun to use AI tools to strengthen their defenses. He expects that similar large-scale attack trends will also spread to the traditional internet industry.
He has previously analyzed DeFi security data, noting that while the number of attacks is rising, the median loss per attack is decreasing. Large protocols become overall safer after implementing AI fortifications, while small and abandoned projects become primary targets; the annualized amount stolen remains within historical ranges.
The on-chain environment, due to its instant monetization and open code characteristics, has become a priority testing and scaling ground for attackers; large protocols transfer defense costs into competitive advantages through investment in AI auditing and fortification, while small protocols are concentratedly plundered due to insufficient maintenance.
Security spending is shifting from traditional manual audits to AI-driven continuous testing, with funds and resources flowing towards large protocols and security tool providers capable of sustained investment; beneficiaries are top protocols and security infrastructure that have completed AI fortifications, while those under pressure are small DeFi projects lacking maintenance and traditional web applications facing similar attack densities.
Source: Public Information
ABAB AI Insight
As a Managing Partner at Dragonfly, Haseeb Qureshi has long tracked DeFi security and the impact of AI on cryptography; he previously analyzed that the annualized amount stolen in DeFi in 2026 would be lower than in 2025, pointing out that attacks are concentrated on small protocols, and noted that AI has reduced the cost of vulnerability discovery to a few dollars (for example, about $2 for AI auditing in the Coldcard case to reproduce vulnerabilities), opposing the "crypto doomsday theory" and advocating for mechanisms like hash backups to address potential cryptographic breakthroughs.
In terms of capital pathways, he promotes shifting security budgets from one-time audits to continuous AI fortification, where large protocols establish barriers through higher security spending, while Dragonfly, as an investor, allocates resources to top projects and security tools that can bear ongoing security costs; the motivation is to use AI to lower marginal defense costs, allowing defenders to gain an advantage in the relevant search space rather than relying on the independence of attackers.
This is similar to the early internet, where the proliferation of vulnerability scanning tools led to mass intrusions of small sites while large enterprises reinforced through continuous investment; it also resembles the financial industry's shift to real-time monitoring and insurance after the emergence of automated attack tools. Currently, we are in the early stage where AI tools are scaling both attacks and defenses, with small projects undergoing mass clearance and large protocols entering a phase of defensive advantage.
Essentially, this is a technological substitution: AI transforms vulnerability discovery and fortification from a high-entropy manual process to a low-entropy scalable process, synchronously reducing defense and attack costs; the reason is that model correlations lead to converging attack ideas, allowing defenders to match only the highest attack budget rather than the total budget to cover major risks, thus shifting security from "covering all possibilities" to "matching the highest investment."
ABAB News · Law of Cognition
The rise in the number of attacks and the decrease in average losses indicate a clearing phase.
Security = Continuous Investment × Model Correlation
On-chain is an early warning indicator for internet security.