
Security researchers at Ikotas Labs, in partnership with OpenAI, have uncovered a severe vulnerability in the widely used NGINX web server, which they have dubbed “NGINX2Shell.” The defect is a pre‑authentication remote code execution flaw that allows an attacker to run arbitrary commands on a target system without needing any credentials or prior access.
The issue is catalogued as CVE‑2026‑123007 and affects NGINX Open Source versions from 1.9.5 through 1.31.8. It has been assigned a CVSS v4.0 severity rating of 9.2, placing it in the “Critical” tier. In addition to remote code execution, the vulnerability can trigger a denial‑of‑service condition when the server’s HTTP/2 feature is enabled.
NGINX version 1.9.5, first released in 2015, marks the earliest affected release, meaning the vulnerable code has been present in production environments for more than a decade. The disclosure was accompanied by a second flaw, CVE‑2026‑123633, which is a use‑after‑free bug in the HTTP/2 implementation. This second vulnerability applies to versions 1.19.7 through 1.31.8 and requires additional conditions to be exploitable, narrowing its practical impact.
Both CVEs were made public on 18 November 2026. On the same day, F5, the company that owns NGINX, issued patched releases—NGINX Open Source 1.31.9 and 1.30.7. Ikotas Labs acknowledged OpenAI’s role in both the discovery and coordinated disclosure in a post on X dated the same day.
The research also revealed that Ikotas was not alone in examining this part of the codebase. Independent analyses by other security firms, including Wiz, Rapid7 and Xint, identified overlapping weaknesses. Rapid7 went further by publishing a functional proof‑of‑concept exploit against an official NGINX 1.31.8 Docker image; the demonstration required roughly 128 GB of traffic in a controlled lab environment. While proof‑of‑concept code is now publicly available, no reports of active exploitation in the wild have surfaced.
NGINX serves as a core web server and reverse‑proxy for a large share of global internet traffic. Because a reverse proxy sits in front of backend services, a compromise can have cascading effects on the entire infrastructure it protects.
Operators of affected NGINX installations are urged to upgrade without delay to the newly released 1.31.9 or 1.30.7 versions. Given that exploit code is already accessible, the interval between disclosure and potential real‑world attacks represents the primary danger. Organizations unable to patch immediately should at least assess whether HTTP/2 is enabled and evaluate the exposure of any vulnerable instances.
The manner in which the flaw was uncovered underscores a broader trend in vulnerability research. The collaboration with OpenAI, along with the parallel discoveries made by several firms using AI‑assisted tools, suggests that artificial‑intelligence technologies are becoming integral to security investigations. For AI research labs, successful disclosures such as this high‑severity bug provide tangible evidence of the practical value their tools can bring to the field.
Source: CryptoNews.net