OpenAI releases GPT-6 Cyber, a specialized model for authorized vulnerability research
GPT-6 Cyber achieves 48.5% on ExploitGym, up from 32.1% for GPT-5.6 Cyber, with improved alignment to stay within authorized scope during security testing.
GPT-6 Cyber achieves 48.5% on ExploitGym, up from 32.1% for GPT-5.6 Cyber, with improved alignment to stay within authorized scope during security testing.
OpenAI's cyber security research identified nearly 1,000 local privilege escalation vulnerabilities in the Linux kernel using AI, scaling vulnerability discovery and validation at scale.
OpenAI is launching a new security sprint with external researchers and quadrupling critical vulnerability rewards as part of an AI-driven defense initiative.
OpenAI raises critical vulnerability payouts to $30,000–$172,000 and introduces tiered loyalty bonuses to attract long-term security researchers.
Research shows agentic AI excels at finding vulnerabilities with human guidance but lacks the conceptual creativity to devise novel attacks autonomously.