AI Exposes Hidden PHP Flaws

The PHP Foundation issued its long-awaited security patches on Tuesday, October 21, bringing the coordinated release forward from its original schedule. This accelerated timeline was a direct response to premature disclosures that occurred the previous day, which triggered widespread public concern. The updates address two critical vulnerabilities that had eluded human detection for nearly two years, marking a significant shift in how cybersecurity threats are identified and resolved in the modern digital landscape.

The scope of the impact is vast. PHP is an open-source language that currently supports an estimated 70% of the internet. The Foundation released specific fixed versions to remedy the issues, including PHP 8.2.35, 8.3.36, 8.4.27, and 8.5.12. Notably, versions 8.1 and older will not receive official corrections. OpenAI, which announced the discovery in a press release the same day, considers this incident one of the most significant cybersecurity findings of 2026. The company attributes the discovery to its GPT-5.6 Sol model, which was utilized under its “Patch the Planet” cybersecurity initiative.

The investigation began in late July at the request of the PHP Foundation, which was already aware of the underlying defect but lacked a method to exploit or fix it. Researchers from Calif, a California-based cybersecurity firm and the program's founding partner, started their audit using OpenAI's tools on July 30. On August 11, Calif researcher Ryan A. identified the first exploitable vulnerability, with OpenAI researchers confirming the finding on the same day. Shortly after, another security firm, Wiz, independently discovered a second, related flaw.

The two vulnerabilities, designated as CVE-2026-109618 and CVE-2026-109619, stem from errors in how PHP manages memory and processes fatal errors. The first issue arises when the system runs out of allocated memory, leaving incomplete data that the engine may later misinterpret as valid. The second involves a fatal error causing the system to point to already-freed memory. Both flaws allow remote attackers to execute malicious code without login credentials, potentially taking control of servers. They also carry a critical severity score of 9.2 out of 10.

The stakes are heightened by the fact that these bugs reside in PHP's core engine. This means websites can be compromised even if their specific application code is secure. Furthermore, the flaws could expose sensitive data stored in server memory, such as passwords, authentication tokens, and API keys. Human developers had examined the code for two years without success, while AI-assisted researchers turned the dormant bug into working exploits within just two weeks.

OpenAI frames this event as a demonstration of its “Defender's Window” concept. This idea suggests there is a narrow period where organizations can use advanced AI to find and patch defenses before those same tools become widely available to malicious actors. The company argues that it is a collective responsibility to act proactively, citing recent findings by Anthropic’s Claude Mythos, which identified thousands of zero-day vulnerabilities across major operating systems and browsers. While some argue the flaws would be harmless without AI, OpenAI maintains that powerful technology will inevitably be used by bad actors, making preemptive action essential.

The lengthy two-month timeline to fix the issues was attributed to the complexity of the task, which required major changes to the PHP engine. The company coordinated a coalition of researchers, maintainers, security providers, and government agencies to ensure a robust solution. For site administrators, the immediate step is to upgrade to the latest patched versions to remain secure.