Technology Company Nvidia released a software update for a patch vulnerability in Triton Server on Saturday. The client uses it for artificial intelligence models.
A vulnerability that cybersecurity company Wiz calls “critical” can lead to AI models acquisitions, data theft and response operations if patches are not applied.
“Wiz Research discovered a chain of vulnerabilities that, when combined, can allow attackers without prior access to have full control over AI servers,” Nir Ohfeld, head of vulnerability research at Wiz, told Cointelegraph.
“Attacks start with a small bug that causes the server to leak small pieces of secret internal data,” he said. “Attackers will use that data to trick one of the server's legitimate features into controlling private system components. This initial scaffolding is everything they need to escalate privileges and achieve a full server acquisition.”
https://www.youtube.com/watch?v=gauqbt6ckes
Triton is an open source inference software designed by NVIDIA to optimize artificial intelligence models.
The full range of customers using Triton is unknown, but it is cited as being adopted by some well-known companies, including Microsoft, Amazon, Oracle, Siemens and American Express. According to a 2021 press release, over 25,000 companies use NVIDIA's AI stack.
An Nvidia spokesman declined to comment, not only mention the company's security bulletins. The disclosed vulnerabilities were assigned the identifiers CVE-2025-23319, CVE-2025-23320 and CVE-2025-23334.
“The most important step is to update to a patch version of the Nvidia Triton Inference Server (version 25.07 or later),” Ohfeld told Cointelegraph. “This directly fixes the entire vulnerability chain.”
Ohfeld has said that for now, “we don't see any evidence that these specific vulnerabilities are being exploited in the wild. However, Nvidia Triton is a very popular and widely used platform for AI workloads.”
Related: 5 Smart Contract Vulnerabilities: How to Identify and Mitigate them
Security vulnerabilities hinder emerging technologies
Security vulnerabilities have hampered emerging technologies in 2025, including cryptography, which exploits led to theft of billions of dollars in digital assets.
According to Hacken, a blockchain security auditor, access flaws and smart contract bugs contributed to the $3.1 billion lost in Crypto Exploits in the first half of 2025.
Meanwhile, some experts say AI agents and quantum computing could pose new cyber threats.
magazine: Internal Pink Drainer – Security Analyst defends his crypto fraud franchise
