Dragonfly: Artificial Intelligence Has Not Met Hackers' Expectations in DeFi
Contrary to gloomy forecasts, a large-scale hacker invasion using artificial intelligence in the DeFi sector did not occur. This is evidenced by data from Hasib Qureshi, managing partner of venture firm Dragonfly. The analyst notes that despite a record number of incidents, the median damage from attacks in 2026 decreased to $500,000, whereas a year earlier this figure stood at $2 million.
Qureshi emphasizes that attackers armed with AI focused on small or abandoned projects. Large DeFi protocols, on the other hand, managed to significantly strengthen their defenses, which prevented major losses. This indicates market maturity: segment leaders recognized the threats and invested in security, while smaller players remain vulnerable.
My expert opinion: The trend of decreasing median damage amid a rising number of attacks is a classic sign of "weeding out" weak links. The DeFi sector is undergoing natural selection, where only projects with robust architecture survive. AI has not yet become a "universal master key" for hacking, but its role in automating phishing and social engineering will grow. Large protocols should not let their guard down, and smaller ones need to consider partnerships with auditors.