Vulnerability Research

Vulnerability research is the work of finding security flaws in software, systems, and AI tools before attackers exploit them. For AI governance teams, it matters because the discoveries often reveal weaknesses in products the organization already uses, such as AI coding assistants, model-serving platforms, or agent frameworks. Tracking this research helps risk managers decide when to patch, restrict access, or pause a deployment.

1 item

ResearchGlobal2026-10-16

A $40K AI Run Cracked Chrome. Offensive AI Risk Is Now a Board Issue.

Security research firm Xbow used OpenAI's GPT-5.6 Sol over six days, spending roughly $40,000, to find a chained vulnerability in Google Chrome affecting over one billion Windows machines. Google patched the flaws and awarded Xbow a $250,000 bug bounty. The episode demonstrates that finding critical software flaws using publicly available AI is now within reach of small, modestly funded teams.

GPT-5.6 OpenAI offensive AI vulnerability research AI-assisted exploitation cybersecurity patch management