Wiz Research uncovered multiple critical vulnerabilities within the core of PHP, the popular general-purpose scripting language that powers an estimated 70% of all websites across the world. These flaws, collectively dubbed PHP2Shell, include memory disclosure and unauthenticated remote code execution within PHP itself, meaning the security risk exists independently of the specific application running on top of the server.
The severity of these flaws is immense, as they allow unauthenticated attackers to potentially execute code on affected servers or read sensitive process memory. This exposed memory could contain critical data such as passwords, authentication tokens, API keys, and certificates belonging to other requests or the server itself. The vulnerabilities were identified as CVE-2026-109618 and CVE-2026-109619.
A striking element of this discovery is the longevity of the flaws; every PHP version released in the last decade is affected by at least one of these vulnerabilities, tracing back to version 5.4. These weaknesses had resided within PHP’s codebase for more than two decades before AI-assisted security research brought them to light, significantly reducing the barrier to exploitability.
The initial discovery was substantially aided by AI tools, which reduced the complexity required to turn a longstanding architectural flaw into a functional exploit. While the flaws were deep-seated, the ability of AI to perform the precise reasoning needed to jump to a specific point in memory and overwrite the correct bits transformed a theoretical issue into a demonstrable threat.
The urgency surrounding this issue was immediate. On October 21, 2026, PHP released the patch addressing these memory-safety vulnerabilities in the core engine, in security releases PHP 8.5.12, 8.4.27, 8.3.36 and 8.2.35, and urged all users to update immediately. This rapid response followed intense activity; Amazon Web Services reported seeing attempts by attackers to use the exploit in the days leading up to the public disclosure on October 20. This swift exploitation prompted multiple authorities to take websites offline as a precautionary measure.
The impact was immediate and global. Shortly after reports broke, exploit attempts surged dramatically, reaching millions by the evening of October 21, 2026. The existence of these flaws, which could allow an attacker to take over any computer running a website without a password, necessitated widespread mitigation efforts across the internet.
Sources: