AI-found bugs aren't proving any easier to exploit despite the hype

Summary

VulnCheck says fewer than 2% of AI-assisted vulnerability discoveries have been weaponized, casting doubt on claims frontier models are handing attackers a major advantage