AI models help uncover chained PHP2Shell remote-code-execution flaws
O’Reilly Radar, citing Wiz Research, reports that Wiz Atlas, Gemini 4 Argon and GPT-5.6 Sol helped uncover two chained PHP remote-code-execution vulnerabilities, later identified as CVE-2026-109618 and CVE-2026-109619. The PHP Foundation released PHP 8.5.12, 8.4.27, 8.3.36 and 8.2.35 and urged all installations to update, while sites unable to upgrade were advised to go offline and be isolated. The reported timeline also says Amazon apparently detected exploitation before the vulnerabilities were prematurely disclosed and patched.
The incident demonstrates that frontier AI models can make complex, long-standing vulnerability chains practically exploitable, while the reported month-long gap before PHP began fixing them highlights the remediation burden on legacy-software maintainers.
This source provides an excerpt. Continue reading on the original site.
#PHP2Shell#PHP#Wiz Research#Wiz Atlas#Gemini 4 Argon#GPT-5.6 Sol#PHP Foundation#Amazon#WordPress#Drupal#Wikimedia