Red Hat Security Advisory: Red Hat Enterprise Linux AI 3.3.6 enhancement update

This advisory covers security vulnerabilities in Red Hat Enterprise Linux AI, specifically the AI Inference Server Model Optimization Tools 3.3.6 (CUDA). Multiple CVEs are addressed, including CVE-2026-34993 and others, involving various weaknesses such as CWE-502 (Deserialization of Untrusted Data), CWE-125 (Out-of-bounds Read), CWE-79 (Cross-site Scripting), and others. The update provides new container images to fix these issues. The vendor has released an official security advisory (RHSA-2026:59518) confirming the availability of these fixes. No exploits are currently known to be active in the wild.

The vulnerabilities could potentially allow attackers to exploit weaknesses in the AI Inference Server Model Optimization Tools, possibly leading to unauthorized actions or system compromise. The exact impact depends on the specific vulnerability but includes risks related to memory corruption, improper input validation, and other security flaws. The severity is rated as high by Red Hat.

An official fix is available via the Red Hat Enterprise Linux AI 3.3.6 update, which includes updated container images addressing the listed vulnerabilities. Users should apply this update promptly to mitigate the risks. No additional mitigation steps are specified or required beyond applying the official update.