AI found more critical bugs in another widely used technology running 225M websites
First reported by The Register ·
Critical vulnerabilities in NGINX are now being discovered by AI, leading to a faster patching cycle and increased risk.
F5 has issued emergency patches for 10 vulnerabilities discovered in the open-source NGINX web server, a technology that powers approximately one-third of all websites worldwide. Two of these vulnerabilities, CVE-2026-123633 and CVE-2026-123007, are rated critical and allow for unauthenticated remote code execution when HTTP/2 is enabled. One of these bugs has existed in NGINX code for at least 10 years, with the affected versions dating back to 2015. Researchers utilizing AI independently discovered these critical flaws, as well as others, which could lead to data theft and service disruption. While F5 has stated there is no evidence of active exploitation, they are monitoring the situation closely and advise users to update to NGINX version 1.31.9 as soon as possible to address these security risks.
The discovery of critical vulnerabilities in NGINX by AI highlights a paradigm shift in cybersecurity. AI is accelerating the identification of long-standing risks in foundational internet technologies, forcing a reevaluation of security postures. This wave of AI-discovered flaws suggests that the speed at which vulnerabilities are found and weaponized may outpace existing defenses, creating a significant challenge for security teams tasked with patching and mitigating these risks.
The implications extend to cloud providers and organizations running large-scale web infrastructure, as demonstrated by the Xint team's "nginx2root" exploit. This exploit chain, leveraging an NGINX vulnerability alongside a Linux kernel flaw, allows for root privilege escalation across major cloud platforms and bare-metal servers. While AI-powered research may eventually harden software, the immediate future presents a "nightmare" scenario for administrators dealing with a deluge of critical CVEs, necessitating enhanced collaboration between researchers, vendors, and defenders to manage the escalating threat landscape.
AI-written summary. May contain errors.