Dragonfly’s Haseeb Qureshi Says AI Vulnerability Scanning Has Become a Cost Race

Dragonfly managing partner Haseeb Qureshi said security has become a cost race between attackers using frontier AI to scan for vulnerabilities and defenders spending on AI-based hardening. According to Odaily, he said Claude found the flaw in eight minutes and GLM reproduced it in 20 minutes, estimating that GLM 5.2’s discovery cost for reproducing the attack was about $2.

Qureshi said this means the vulnerability could be found with AI for about $2. He also said small companies will struggle to survive in security-sensitive product areas and that any such product should undergo an AI security scan before each release.