Haseeb Claims DeFi Has Not Experienced a 'Hacker Apocalypse'
Haseeb stated that with the launch of models like GLM 5.2, Fable, and GPT 5.6, which have been utilized by attackers, DeFi has not encountered the anticipated 'hacker apocalypse'.
Based on current data and rates, the annualized amount stolen from DeFi in 2026 is approximately $1.89 billion, with a total of about $986 million stolen within the year, which is lower than the levels of 2025 and remains within historical ranges.
Market Mechanism: The number of attacks has increased, but the scale of each attack has decreased more rapidly. Hackers are shifting towards smaller protocols and abandoned projects, while the security of large protocols has been reinforced, maintaining the safety of funds. Overall, market funds are flowing towards large DeFi protocols that have strengthened security, benefiting compliant large protocols while putting pressure on smaller projects.
Source: Public Information
ABAB AI Insight
As a managing partner at Dragonfly, Haseeb has previously analyzed trends in crypto security and has been cautiously optimistic about the impact of AI models on code audits. This time, he directly refutes the 'hackpocalypse' prediction with data, continuing his data-driven style.
In terms of capital pathways, funds like Dragonfly are mobilizing resources by supporting security tools and audit projects. Large protocols are increasing investments in AI hardening to counter new model attacks, motivated by the need to protect TVL and maintain DeFi's status as a mainstream capital entry point.
Similar to the industry's shift towards audits and insurance following multiple cross-chain bridge attacks in 2022-2023, DeFi is currently transitioning from vulnerable small projects to a concentration of security in large protocols.
Structural Judgment: This essentially belongs to a capital concentration driven by technological substitution. While new AI models enhance attack efficiency, they also accelerate protocol security upgrades. Attackers are shifting towards smaller targets, leading to a concentration of funds in large protocols that can afford AI hardening. The mechanism is that projects with high safety margins attract more TVL and liquidity, creating a positive cycle.
ABAB News · Law of Cognition
The stronger the attack tools, the faster the defense iterations; security becomes a winner-takes-all arms race.
As the number of attacks rises, the scale decreases, with risks shifting from being dispersed to concentrated among the weaker.
Large protocols use AI for self-protection, while small projects become cannon fodder; capital always chases after moats.