OpenAI says new GPT-5.5-Cyber outperforms Anthropic's Mythos on cybersecurity benchmark

2026-06-23 · Source: The Decoder · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Cybersecurity & Data Privacy · Depth: Intermediate, short

Summary

OpenAI has significantly expanded its Daybreak cybersecurity initiative, introducing an updated Codex Security plugin and fully releasing the specialized GPT-5.5-Cyber model. The Codex Security plugin, which previously scanned over 30 million commits across 30,000 codebases, now handles the entire vulnerability workflow from discovery to patch generation, automatically flagging 500,000 fixes and confirming 70,000 with human review. GPT-5.5-Cyber, designed for finding and patching software flaws, outperforms competitors on key benchmarks, achieving 85.6% on CyberGym, 39.5% on ExploitGym, and 69.8% on SEC-bench Pro, surpassing Anthropic's Mythos 5 (83.8% on CyberGym). Access to GPT-5.5-Cyber is restricted to verified defenders. OpenAI is also collaborating with over 25 security firms and several governments through a dedicated partner program and launched "Patch the Planet" to bring patching tools to over 30 open-source projects.

Key takeaway

For AI Security Engineers evaluating vulnerability management solutions, OpenAI's expanded Daybreak initiative offers a significant shift towards automated remediation. You should consider integrating the updated Codex Security plugin into your development pipeline to automate patch generation and triage findings. Explore the Daybreak Cyber Partner Program for access to GPT-5.5-Cyber, which can enhance your team's ability to discover and fix software flaws more efficiently.

Key insights

OpenAI's Daybreak initiative shifts cybersecurity focus from vulnerability discovery to automated patching and remediation.

Principles

• Automated patching closes the security bottleneck.
• Specialized AI models enhance security benchmark performance.
• Restrict advanced AI access to vetted defenders.

Method

The Codex Security plugin analyzes code, spots flaws, checks reachability, builds targeted patches, and verifies results, integrating with existing vulnerability management systems.

In practice

• Integrate Codex Security for automated patch generation.
• Utilize GPT-5.5-Cyber for enhanced vulnerability discovery.
• Join Daybreak partner program for advanced AI access.

Topics

• Cybersecurity Automation
• Vulnerability Management
• GPT-5.5-Cyber
• Codex Security Plugin
• AI Security Benchmarks
• Open-Source Security

Best for: CTO, VP of Engineering/Data, Investor, AI Security Engineer, AI Engineer, Director of AI/ML

Related on AIssential

• Daybreak: Tools for securing every organization in the world — AI has shifted the cybersecurity bottleneck from vulnerability discovery to patching, necessitating automated remediation at machine speed.
• OpenAI Launches Full-Scale Effort to Patch Open-Source Bugs as It Takes on Anthropic’s Mythos — OpenAI is significantly expanding its cybersecurity initiatives and tools to address AI hacking risks.
• OpenAI just released its answer to Claude Mythos — OpenAI's Daybreak uses AI to proactively identify and patch software vulnerabilities.
• OpenAI Is Sending Security Engineers to Fix the Internet’s Most Neglected Code — AI can effectively augment human security engineers to proactively identify and fix critical open-source vulnerabilities.
• OpenAI expands Trusted Access program with GPT-5.5-Cyber — OpenAI is deploying a specialized AI model, GPT-5.5-Cyber, to enhance cybersecurity defenses for critical infrastructure.
• Introducing Patch the Planet — AI models like GPT-5.5-Cyber accelerate vulnerability discovery, shifting the security challenge to patching and long-term code hardening.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by The Decoder.