# OpenAI Launches $1 Billion 'Patch the Planet' Cybersecurity Initiative

> OpenAI has launched the Patch the Planet initiative to uncover software vulnerabilities, following the discovery of a critical PHP flaw using GPT-5.6 Sol.

- Canonical URL: https://coreiten.com/en/article/openai-patch-the-planet-initiative-targets-critical-flaws
- Language: en
- Section: Tech News
- Author: Sami
- Published: 2026-10-22T03:10:59+03:00
- Modified: 2026-10-22T03:10:59+03:00
- Publisher: CoreITen (https://coreiten.com)
- Keywords: OpenAI, Patch the Planet, PHP2Shell, GPT-5.6 Sol, Daybreak, GPT-6 Astra, Anthropic Mythos

## Summary

OpenAI launched a $1 billion initiative called 'Patch the Planet' to secure aging software after discovering critical PHP flaws using AI.

- The emergency patch deployed on October 21 targets two major vulnerabilities known as PHP2Shell, which affect PHP technologies powering up to 70 percent of the world's websites.
- OpenAI's GPT-5.6 Sol model successfully identified the PHP2Shell flaws, which the company described as one of the most impactful security issues in recent memory.
- To support resource-limited organizations defending essential services, OpenAI is dedicating $1 billion to subsidize access to its Daybreak cybersecurity auditing platform.
- OpenAI reported that newer models like GPT-6 Astra and GPT-5.6 Sol have shown substantial improvements in translating legacy codebases such as COBOL and Fortran.

**Why it matters:** This initiative highlights both the immense defensive power of advanced AI in securing legacy systems and the escalating threat of AI-driven cyberattacks.

---

OpenAI has introduced a new cybersecurity framework called "Patch the Planet" aimed at fostering global cooperation to find and fix vulnerabilities in older software. The announcement follows the release of an emergency patch on October 21 for PHP, a technology that powers up to 70 percent of the world's websites.

### The PHP2Shell Vulnerability

The emergency patch addresses two major vulnerabilities, collectively known as PHP2Shell, which could allow hackers to take control of web servers if exploited. According to reports, these flaws were discovered using OpenAI's GPT-5.6 Sol model, demonstrating the ability of modern AI to identify complex security issues that humans might miss.

In a [blog post](https://openai.com/index/coordinated-disclosure-of-critical-vulnerabilities/), OpenAI described PHP2Shell as "one of the most impactful security issues in recent memory." The company noted that while the PHP flaws are particularly dangerous due to the sheer number of potential targets, they are just two among several significant vulnerabilities it has already found.

### Funding and Tools for Cyber Defense

To support organizations with limited security resources that protect essential services, OpenAI is dedicating $1 billion to subsidize access to Daybreak, its cybersecurity auditing platform. The company has also released a [Codex skill on GitHub](https://github.com/openai/php2shell-checker) designed to help users check if their websites have been compromised by PHP2Shell.

The "Patch the Planet" initiative operates as a crowdsourced mission, inviting third-party cybersecurity professionals to review OpenAI's approach, challenge its assumptions, and identify gaps. In its [announcement](https://openai.com/index/coordinated-disclosure-of-critical-vulnerabilities/), OpenAI stated the project aims "to help turn findings into protection and build a response that can scale as AI capabilities advance."

### Modernizing Legacy Systems

Many essential services, including banks and public sector infrastructure, still rely on decades-old programming languages like COBOL and Fortran. In a separate blog post published on October 19, OpenAI reported that its latest models, including GPT-6 Astra and GPT-5.6 Sol, have shown "substantial improvements" in translating codebases from one programming language to another.

### The Dual Nature of AI in Cybersecurity

The cybersecurity sector has faced a crisis in recent months as powerful new AI systems, such as Anthropic's Mythos, threaten to render existing cyberdefense infrastructure obsolete. OpenAI acknowledges this double-edged sword, stating that while AI can uncover internet-scale security flaws, "the same capabilities can also help attackers rapidly develop offensive tools and exploits." Moving forward, the company expects the discovery of similar vulnerabilities to increase sharply in the coming months.

## Sources

- [io9.gizmodo.com](https://gizmodo.com/openai-wants-to-patch-the-planet-with-its-own-ai-2000829119)

## Related topics

- [OpenAI](https://coreiten.com/en/topic/openai)
- [GPT-5.6 Sol](https://coreiten.com/en/topic/gpt-56-sol)
- [GPT-6 Astra](https://coreiten.com/en/topic/gpt-6-astra)