OpenAI Launches GPT-5.5-Cyber and Patch the Planet to Automate Open-Source Vulnerability Fixes
Summary: OpenAI expanded its Daybreak cybersecurity program on June 22 with the full release of GPT-5.5-Cyber and Patch the Planet — a program designed to close the gap between finding vulnerabilities and actually fixing them across open-source infrastructure.
Key Facts
- GPT-5.5-Cyber benchmarks: CyberGym 85.6% (vs. 81.8% for base GPT-5.5), ExploitGym 39.5% (+13.5 pp), SEC-bench Pro 69.8% — top scores across all three defensive security evals
- Patch the Planet: Trail of Bits dedicated its full security research team to 19 open-source projects for a five-day sprint, already merging dozens of patches with coordinated disclosure pending for more
- Linux Kernel alone: 8 kernel pointer information-leak PoCs and 24 local privilege-escalation exploits identified
- 30+ security partners: Cloudflare, CrowdStrike, Palo Alto Networks, Wiz, and others — suggesting broad enterprise adoption is planned
Why It Matters
The bottleneck in cybersecurity is shifting: AI can now find vulnerabilities faster than humans can patch them. Patch the Planet directly targets that gap, automating the full remediation loop — discovery, validation, patching, and deployment — across the open-source infrastructure that underpins nearly all modern software. If it scales, it could raise the baseline security of shared infrastructure in a way that point-in-time audits never could.
Read More
- Patch the Planet announcement — OpenAI
- GPT-5.5-Cyber deep dive — The Hacker News
- Daybreak full expansion — SiliconANGLE