Two critical, unauthenticated remote-code-execution flaws affect NGINX’s handling of HTTP/2: CVE-2026-123633, a use-after-free, and CVE-2026-123007, an integer underflow that can cause an invalid memory copy. Both are rated CVSS 9.2 and require HTTP/2 to be enabled. The first affects reverse-proxy processing when available connection slots run low; the second can also affect servers that serve content directly. Testing showed the latter could compromise a server serving ordinary static files, without a backend application or third-party modules. An attacker could run commands, steal data, alter websites, or disrupt services, potentially affecting multiple applications behind one NGINX deployment. The older flaw dates to version 1.9.5, released in 2015.
The vulnerabilities lie in NGINX itself, so secure application code does not prevent exposure. Fixes are available in NGINX OSS 1.30.7 stable and 1.31.9 mainline, and in NGINX Plus 37.0.8.1 and 37.3.0.1; organizations should identify affected deployments and upgrade promptly. If upgrading is delayed, temporarily disable HTTP/2, test compatibility, reload NGINX, and verify that old worker processes have exited. Unexpected worker crashes or commands launched by NGINX warrant investigation, though they are not confirmed indicators of exploitation. The findings also illustrate a broader security shift: AI-assisted analysis is helping researchers uncover subtle memory-safety defects that remained hidden for years, while shortening the time defenders may have to deploy fixes and reduce exposure.
Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.