Global – October 31, 2026 -- A PHP vulnerability that stalled in triage after a 2024 audit became unauthenticated remote code execution this year once researchers applied frontier AI models, and both PHP2Shell CVEs now carry a CVSS v4.0 score of 9.2.

PHP Foundation admits a 2024 finding stalled for lack of a proven exploit

A 2024 external audit led a PHP core developer to the first signs of an out-of-memory problem, according to the PHP Foundation's account of CVE-2026-109618. An exploit was theorized, but no one found a concrete attack vector. Each candidate fix carried a cost: a compatibility break, new denial-of-service potential, or changes to third-party extensions that seemed impossible to coordinate. Work on the fix stalled and restarted for months.

Calif, auditing PHP through OpenAI's Patch the Planet program, then reported an unconditional remote code execution with the same root cause. OpenAI reported the same root cause independently, and Wiz added more attack paths. On Apache with mod_php, an unauthenticated request can trigger the bug before any application code runs. The PHP Foundation called the delay a structural failure of its own process and resources.

PHP2Shell needs no prior access, unlike four AI-found Linux kernel flaws

Copy Fail, Dirty Frag, Fragnesia, and CIFSwitch hit the Linux kernel in April and May. Each required an attacker who already had system access to escalate privileges. PHP2Shell arrives over HTTP against the runtime web applications depend on.

Nathan Blackham, who leads the CIQ engineers who shipped those kernel fixes, expects AI-found CVEs next in other language runtimes and in libraries every distribution ships but few people read closely.

Google counts monthly disclosures rising from 5,045 to 10,740 in eight months

Google's threat intelligence group reports that monthly vulnerability disclosures rose from 5,045 in January 2026 to 10,740 in August. Exploitation in the wild averaged 18 vulnerabilities a month in 2026, up from 10.5 in 2025. From January to August 2026, Google recorded 141 vulnerabilities that were both disclosed and exploited, more than the 127 recorded for all of 2025.

Volume alone does not measure risk. Disclosures naming the Linux kernel accounted for about 5,000 entries over those eight months, with no observed in-the-wild zero-day exploitation. Only about 0.23% of 2026 disclosures were seen exploited. Yet 50% of the vulnerabilities Google identified as likely AI-discovered lead to remote code execution, against 26% across the broader set.

Triage must rank findings by consequence, not by proven exploitability

Google's report places AI models' strongest results in memory corruption and logic bypasses in core C and C++ libraries, language runtimes, and hypervisors. For those components, CIQ advises ranking a finding by what it would allow if the exploit were proven, and by how exposed the affected systems are. Internet-facing PHP servers running mod_php topped that ranking for PHP2Shell.

Upstream patches cover PHP 8.2 to 8.5 only, leaving enterprise distributions to backport

The PHP Foundation said AI tools find bugs faster for defenders and attackers alike, but cannot weigh "compatibility trade-offs" or review the patch. The PHP project released fixes for supported branches 8.2 through 8.5. For 7.4, 8.0, and 8.1 it published source-only fix branches with no official binaries, so distribution engineers must build, test, and ship those packages. PHP 7.2 received no upstream fix.

Rocky Linux 9 ships PHP 8.0, and Rocky Linux 8 ships PHP as module streams including 7.2, 7.4, and 8.0. CIQ engineers shipped fixed base PHP 8.0 packages for Rocky Linux from CIQ Pro (RLC Pro) LTS 9.6 and 9.2 about three and a half hours after the upstream fix went public. Fixes for several module streams, and for CIQ Bridge on CentOS 7.9, followed over the next three days. A few older streams will not get a fix, including php:7.2, the default stream on Enterprise Linux 8.