Web infrastructure worldwide has faced a coordinated security mobilization following the disclosure of two critical vulnerabilities affecting the PHP runtime environment. Tracked as CVE-2026-109618 and CVE-2026-109619 and collectively known as "php2shell," these severe security flaws impact all web servers running PHP 5.4 or later. Because PHP underpins nearly 70 percent of all websites on the internet—including major content management systems and enterprise frameworks like WordPress and Magento—the vulnerabilities carry immense implications for the broader digital ecosystem.
The security flaws were publicly disclosed following a coordinated effort by security organizations including Calif.io, Wiz, OpenAI, and Quarkslab. The discovery prompted a rapid response from the PHP Foundation alongside major infrastructure and security partners to deliver emergency mitigations, secure runtime updates, and defensive protections before malicious exploitation could scale widely across the web.
Understanding the php2shell Flaws: Memory Corruption and Unsafe Cleanup
The php2shell vulnerabilities stem from how the PHP runtime manages internal memory during error handling and unexpected disruptions in HTTP request processing. Under normal operational circumstances, PHP allocates specific memory resources to process an incoming web request and subsequently releases those resources once the computational work is successfully completed. However, when processing is abruptly interrupted—such as during an out-of-memory condition or a fatal runtime error—PHP fails to clean up memory allocations safely and predictably.
The two distinct CVEs represent different phases and mechanisms of this unsafe memory handling. The first vulnerability, CVE-2026-109618, occurs when PHP runs out of memory and halts its operations before internal data structures have been fully and properly initialized. During the subsequent cleanup phase for that interrupted request, the PHP runtime mistakenly treats this unfinished, partially initialized data as though it were completely valid, opening the door to memory corruption.
The second vulnerability, CVE-2026-109619, involves scenarios where a fatal error interrupts a request, leaving the PHP engine holding active references to memory blocks that are scheduled for release when the request terminates. When the same underlying PHP process handles a subsequent, unrelated request at a later time, it may inadvertently follow those outdated references. This results in the process accessing memory that is no longer valid, potentially exposing sensitive customer data or executing arbitrary code supplied by an attacker.
Because these flaws reside deep within the core PHP runtime architecture rather than in application source code, updating individual website scripts, themes, or plugins provides no remediation. Website administrators and hosting providers must apply the official PHP software updates directly to their servers to permanently resolve the underlying code defects.

Industry-Wide Mitigations and Early Detection
Ahead of the formal public disclosure, threat intelligence and infrastructure networks observed active exploitation attempts. Attackers were actively probing applications and mutating malicious payloads in real time to test defenses and bypass traditional signature-based rules.
In response to leaks and active exploitation attempts, security providers mobilized quickly. Cloudflare, working in coordination with the PHP Foundation, OpenAI, Wiz, and other industry stakeholders, developed and deployed specialized Web Application Firewall (WAF) protections well ahead of the official release date. Automated rules were rolled out to protect customers using free and paid WAF tiers, ensuring that sites were shielded against known php2shell exploit attempts by default.
Protecting against sophisticated runtime memory corruption vulnerabilities required moving beyond standard request-based security rules. Traditional analysis of incoming HTTP request data alone often lacks sufficient context to reliably distinguish between a genuine exploit attempt and legitimate application traffic, creating a high risk of false positives. To solve this challenge, security engineers developed full transaction detection capabilities. This advanced approach combines inspection signals from both the incoming request and the outgoing response within the same HTTP transaction. By evaluating the server’s response context, security systems can accurately verify whether an exploit attempt succeeded and block malicious responses at the edge before they ever reach the requester.
Ongoing Updates and Additional Session Upload Progress Advisory
As patch deployment continued across the web hosting ecosystem, the PHP project published an informational advisory regarding a related vulnerability, identified as CVE-2026-115560. This newly disclosed flaw impacts PHP’s session upload progress feature and affects fully patched PHP releases, including version 8.5.12. The issue is slated for permanent resolution in PHP’s next regular software release cycle.
Cloudflare confirmed that all WAF customers utilizing PHP’s default upload progress field name are automatically protected against known exploit attempts targeting this secondary vector via existing rule logic, specifically through rules designed for CVE-2026-109618 and CVE-2026-109619. However, administrators who have customized their session.upload_progress.name configuration are not covered by the default rule and must exercise additional caution.
Security analysts continue to monitor global threat telemetry closely, updating managed rulesets as new attack variations, evasion techniques, and exploit mutations emerge. Administrators are urged to maintain their WAF protection layers while systematically applying official vendor patches to all PHP repositories and server environments as quickly as possible.