Dragonfly: Concerns about the "AI tsunami" in DeFi have been exaggerated
Fears that artificial intelligence would trigger an avalanche of hacks in the decentralized finance (DeFi) sector have proven unfounded. Haseeb Qureshi, managing partner at venture firm Dragonfly, has presented a fresh analysis of the situation that debunks the darkest predictions.
The key takeaway: despite a record number of incidents in 2026, the median damage from attacks has sharply declined. While the average loss was around $2 million a year earlier, this figure has now dropped below the $500,000 mark. This is direct evidence that a mass "AI apocalypse" for DeFi has not materialized.
Where are AI hackers striking?
According to Qureshi's observations, attackers using AI-based tools have shifted to less protected targets. Today, the main attacks are directed at small, forgotten, or poorly maintained projects where security levels leave much to be desired. In contrast, major DeFi protocols have significantly strengthened their defenses over the past year, making them unattractive for automated attacks.
This trend confirms that the threat from AI has not disappeared, but it has transformed. Instead of a total collapse, we are seeing a redistribution of risks: peripheral and undercapitalized projects remain vulnerable, while DeFi "blue chips" demonstrate growing resilience.
My expert conclusion: The DeFi market is undergoing a natural stage of maturity. AI has not become an "industry killer" but has merely accelerated the process of natural selection. Projects with good architecture and attention to security survive, while shoddy and abandoned ones fade into oblivion, falling victim to automated algorithms. Investors should focus not on loud headlines but on the fundamental security metrics of protocols, especially in the era of AI.