A long-known flaw can become an urgent service problem when someone finds a practical way to use it. The work of finding that path still matters as much as the age of the bug.
WIRED’s October 24 investigation describes how AI-assisted research helped turn longstanding PHP weaknesses into working exploits, alongside specialist human knowledge. It also reports widespread exploitation attempts while distinguishing them from confirmed successful compromises. That distinction matters: traffic aimed at a flaw establishes pressure on defenders, not a count of breached organizations.
The patches predate this report. The PHP Foundation’s October 21 account identifies fixed releases as PHP 8.5.12, 8.4.27, 8.3.36 and 8.2.35. It describes two critical vulnerabilities, CVE-2026-109618 and CVE-2026-109619, that can allow unauthenticated code execution under the relevant conditions. Application and server configuration affect the reachable paths.
The foundation also explains why remediation took time. Earlier work had identified signs of the memory problem, but a concrete attack path was missing. Possible fixes involved compatibility, performance and ecosystem tradeoffs. It attributes the delay to decision-making and resource constraints, and describes changes to how difficult security decisions reach leadership.
For an enterprise, that turns a security headline into two pieces of work. Confirm the PHP runtime serving each exposed application, including vendor-managed services, and obtain evidence of the applicable update. A changed application package alone may leave the underlying runtime unresolved.
Then examine who can settle a difficult remediation decision. Faster discovery increases the value of experienced maintainers, reproducible reports and someone empowered to choose among imperfect fixes. The immediate deliverable is evidence that the affected service runs the corrected runtime. The lasting one is a decision path that keeps the next hard fix moving.