Dragonfly research shows AI models can identify wallet vulnerabilities for minimal expense, reshaping security economics in crypto.
A Coldcard hardware wallet flaw was discovered using AI for roughly USD 2, according to Dragonfly Managing Partner Haseeb Qureshi. The finding highlights how AI-powered testing is reducing the cost of identifying security vulnerabilities in crypto products.
Qureshi introduced a “Cost of Discovery” metric to quantify the expense of reproducing flaws with frontier AI models. A separate offline test replicated the Coldcard issue in about 20 minutes, while an earlier eight-minute discovery may have relied on web search data.
The shift suggests companies must adapt security strategies as AI lowers the financial barrier to uncovering vulnerabilities, potentially altering risk assessment in the sector.