Ai-driven security collaboration addresses critical PHP vulnerabilities

In October 2026, a global initiative led by AI-driven security teams identified and addressed critical vulnerabilities in PHP, an open-source language used by millions of websites. The flaws, discovered using frontier AI models like GPT-5.6 Sol, were independently confirmed by Wiz and other partners. PHP’s maintainers released patches, with support from Cloudflare, Wiz, and other security providers. A PHP version checker tool was shared to help users apply updates. The vulnerabilities, considered one of the most impactful in recent memory, highlight AI’s dual role in both uncovering and potentially exploiting security risks. The initiative, Patch the Planet, also committed $1 billion in subsidized access to Daybreak for frontline defenders, focusing on organizations with limited resources. A proposed framework for responding to critical vulnerabilities was shared, emphasizing coordination between maintainers, researchers, and government agencies like CISA and the ONCD. The framework outlines five stages for addressing vulnerabilities, from validating fixes to deploying patches and publishing guidance. It also addresses challenges in coordinating global responses and balancing the risks of sharing sensitive details with the benefits of early protection. The initiative aims to improve software security by integrating AI into development processes, helping engineers build more secure systems from the start.