Two PHP flaws, and no fix but the patch
The PHP project on Wednesday shipped four releases fixing two critical flaws that allow unauthenticated remote code execution, each rated 9.2. Wiz Research, one of the teams that reported them, says frontier AI models helped it find the bugs and build working exploits.
PHP's maintainers on Wednesday released PHP 8.5.12, 8.4.27, 8.3.36 and 8.2.35 to fix two critical flaws that can let an unauthenticated attacker run code on a web server, according to the project's security advisory. "There are no complete mitigations, updating is the only safe choice," each release note says.
The flaws are CVE-2026-109618, remote code execution through out-of-memory corruption, and CVE-2026-109619, remote code execution through stale global state. Both are rated 9.2 under version 4.0 of the Common Vulnerability Scoring System, according to the project's GitHub advisories and the US National Vulnerability Database. The project gave advance notice on Tuesday, scheduling the release for 14:00 UTC on Wednesday.
In the first flaw, PHP's handling of an out-of-memory error can interrupt a request while internal data is only partly set up, and the clean-up that follows may then treat that invalid data as valid. In the second, the advisory says, state belonging to one request can survive after its memory is freed, and a later request on the same worker may use it before normal initialisation resets it. The advisory names PHP-FPM, Apache with mod_php, php-cgi, FrankenPHP and LiteSpeed as exposed, and puts the risk highest for mod_php, citing "a demonstrated exploit where an unauthenticated request can trigger the attack before any application code runs."
The stop-gaps are partial. For the memory flaw, the project suggests removing PHP's memory limit in a way applications cannot override and lowering request-size limits, including at the web server or proxy. For the stale-state flaw, it suggests disabling session upload progress and restarting every worker, and warns against relying on ini_set() or .user.ini files. These measures "reduce exposure to specific attack paths but do not make an installation safe," it says.
Support is uneven. Linux distributions and long-term-support vendors received the patches in advance, the advisory says. PHP 8.1 and older no longer receive updates from the project, though its security team published minimal source backports for 7.4, 8.0 and 8.1. The 8.2 branch reaches end of life on 31 December 2026.
Among those credited with reporting the bugs is Wiz Research, the research arm of the cloud security company Wiz, which published its own account on Wednesday. Wiz said it found the first flaw on 19 August "using Wiz Atlas, Gemini 4 Argon, and GPT-5.6 Sol", confirmed unauthenticated code execution against Drupal and WordPress within two days, and reported to PHP on 24 August. "Frontier AI models substantially reduced that barrier, turning a longstanding flaw into an exploit," it wrote, adding that the vulnerable code had been present for more than a decade. Wiz credited the PHP team, the security firm Calif and OpenAI for work on discovery, coordination and patch verification. Wiz's claim that PHP runs 70 per cent of websites is its own.
It was a crowded day. TechCrunch reported that Node.js patched four high-severity flaws, one allowing remote code execution, and that the LiteLLM proxy project said an attacker had published malicious packages to PyPI, Docker Hub and GitHub's container registry.
Why it matters: a pre-authentication code-execution bug in the language beneath a large share of the web, with no configuration that fully closes it, leaves patching as the only control. Wiz's timeline also shows how far AI tools compress the work: one day, by its account, from the first bug to a working exploit against Drupal. As we reported on Wednesday, eight certificate authority keys were the week's other infrastructure scare, and Railway has told every customer to rotate its keys.