Loading vulnerability…
CVE-2025-23319 is an out-of-bounds write vulnerability in the Python backend of NVIDIA Triton Inference Server affecting Windows and Linux deployments. The provided content states that the flaw can be triggered by sending a request to the server and that exploitation is associated with insufficient validation and bounds checking around Python backend/shared-memory handling. In the broader attack chain described in the source material, an attacker can abuse Triton’s shared memory registration path to gain unauthorized read/write access to Python backend private memory and manipulate internal data structures and control flows. Successful exploitation may result in memory corruption and can be leveraged toward remote code execution, denial of service, data tampering, or information disclosure. The content also indicates this issue is especially dangerous when chained with related Triton flaws such as CVE-2025-23320 and CVE-2025-23334.
Mallory correlates every CVE against your assets, your vendors, and active adversary campaigns. Know which vulnerabilities matter for you, not just which ones are loud.
What it means. What to do now. Patch path, mitigations, and the assume-compromise checklist.
What an attacker gets, and what they’ve been doing with it.
If you can’t patch tonight, do this now.
Patch, then assume compromise.
No public exploits tracked yet. Mallory keeps watching.
No public exploit code observed for this vulnerability.
Products and vendors Mallory has correlated with this vulnerability. Open in Mallory to drill down to specific CPE configurations and version ranges.
Vendor-confirmed product mapping. Mallory continuously reconciles this list against your asset inventory.
18 sources tracked across advisories, community write-ups, and news. New activity surfaces here as Mallory finds it.
A vulnerability chain in NVIDIA Triton Server, including CVE-2025-23319, can allow remote takeover of AI systems.
A related NVIDIA Triton Inference Server vulnerability that can be chained with other flaws to enable full unauthenticated remote compromise of Triton instances.
A critical vulnerability in NVIDIA Triton Inference Server mentioned as part of a group of 2025 flaws, but not described in detail in the content.
An out-of-bounds write vulnerability in the Python backend of NVIDIA Triton Inference Server that, when chained with CVE-2025-23320, can enable unauthenticated remote code execution, denial of service, or information disclosure.
Query your assets running an affected version, and investigate the blast radius.
Every observed campaign linking this CVE to a named adversary.
Malware families riding this exploit, with evidence and IOCs.
YARA, Sigma, Snort, and vendor rules, auto-deployed to your SIEM.
Cross-references every affected SKU, including bundled OEM variants.
Community discussion across Reddit, Mastodon, and other social sources.