AI-Assisted Discovery of Critical PHP RCE Vulnerabilities Forces Urgent Global Patching

PHP Discloses Two Critical Vulnerabilities Allowing Unauthenticated Server Takeover; AI Assisted in Discovery, Urgent Upgrade Required for All PHP Sites

On October 21, the official PHP team released emergency updates for versions 8.5.12, 8.4.27, 8.3.36, and 8.2.35 to patch two severe vulnerabilities (CVE-2026-109618 and CVE-2026-109619), collectively referred to by researchers as "PHP2Shell." According to PHP officials, attackers can potentially execute arbitrary code on servers without authentication by sending specially crafted requests, effectively gaining full control of the server. Since WordPress, Drupal, and Magento all run on PHP, OpenAI cites estimates suggesting that approximately 70% of websites rely on it.

[Is My PHP Website Affected?]

Almost certainly yes. The PHP official team states that any PHP installation handling external web requests is at risk. This includes common deployment methods such as PHP-FPM, Apache mod_php, and LiteSpeed, with mod_php posing the highest risk. According to Wiz and Sansec, all versions since PHP 5.4, released in 2012, are affected.

The vulnerability lies within the PHP engine itself—much like a broken lock on a building's main door, no amount of interior security measures will help if the entry point is compromised. Wiz and Calif have already demonstrated unauthenticated attacks against the latest versions of WordPress, Drupal, Laravel, and other frameworks.

[What Should You Do Now?]

• For those managing their own servers: Immediately upgrade to one of the four patched versions mentioned above. Official support for PHP 8.1 and earlier has ended; source patches are only provided for versions 7.4, 8.0, and 8.1. If you use PHP bundled with distributions like Ubuntu or Debian, check your distribution's security updates.
• For standard site owners using shared hosting or managed WordPress: PHP is managed by your hosting provider. Installing plugins or modifying website code will not fix this issue. Check the PHP version in your control panel or directly ask your host if they have applied the patch.

Temporary configuration changes suggested by PHP officials only reduce risk; the official stance is that upgrading is the only complete solution. Cloudflare has implemented blocking rules for all users (including free tier), but they explicitly state this is a temporary measure that only protects traffic routed through Cloudflare. Attackers bypassing Cloudflare to connect directly to the server remain unblocked.

[How Did AI Contribute?]

OpenAI stated that security firm Calif discovered the vulnerabilities using GPT-5.6 Sol as part of its "Patch the Planet" initiative. According to a timeline compiled by Sansec, Calif reported the issues to PHP on August 11. Wiz also independently discovered and reported the flaws, utilizing tools including Gemini 4 Argon and GPT-5.6 Sol. OpenAI claims to have led coordination among maintainers and security vendors over the past two months to prepare fixes, considering this one of the most impactful vulnerabilities in recent years and an example of AI identifying security bugs ahead of attackers.

[Was There a Premature Leak?]

Yes. The vulnerabilities were originally scheduled for public disclosure on October 22. However, someone posted details prematurely on Hacker News on October 20. Although the post was deleted, screenshots had already circulated, prompting PHP officials to move the release date up to October 21.

Attacks have already begun. Cloudflare's CTO noted seeing early attack attempts hours before the patch release. Security firm watchTowr reported that attackers are already using AI-assisted tools to launch exploits, stating, "No one has succeeded yet, but it won't be long." Dutch government sites, including DigiD, proactively went offline to mitigate risk.

PHP Official Announcement: https://t.co/UZYhjf3CWO OpenAI Original Post: https://t.co/jKdbrEaMB8 Wiz Technical Details: https://t.co/SrODSYiXOv

↳ Quoting @OpenAI (OpenAI): OpenAI models helped researchers—operating as part of our Patch the Planet initiative—discover critical security flaws in PHP, an open-source programming language used to run an estimated 70% of websites.

Over the past two months, we led a coalition of maintainers, researchers, and security providers to prepare fixes and protections ahead of disclosure. This duration was necessary because the fix was quite intricate and required a major change to the PHP engine. Every site using PHP can now safely upgrade their version to stay secure, in addition to applying mitigations from various security vendors.

Due to the severity of these flaws and PHP’s widespread use, we consider this bug to be one of the most impactful vulnerabilities in recent memory. It also marks a new phase, where AI can find security bugs with internet-scale impact.

This is an example of the Defender’s Window in action: to our knowledge, frontier AI was able to find this vulnerability ahead of attackers, giving defenders time to prepare.