🧠 Nvidia released an urgent security update on Saturday to address critical vulnerabilities in the open-source Triton server, which is widely used to run AI models.
👨💻 Wiz revealed a series of vulnerabilities (CVE-2025-23319, CVE-2025-23320, CVE-2025-23334) that could allow an attacker to gain full control of the server, steal data, or manipulate the outputs of intelligent models.
> 🗣️ "The attack starts from a simple flaw that leads to the leakage of sensitive internal data. The attacker exploits this data to deceive the system and gain control over a specific component, allowing them to elevate their privileges and gain complete control of the server," according to Nir Ohfeld from Wiz.
💡 Who uses Triton? The Triton server is relied upon by major companies such as:
Microsoft
Amazon
Oracle
Siemens
American Express
According to 2021 data, there are over 25,000 companies relying on Nvidia's AI solutions.
⚠️ What is the required action?
> 🔄 Updating the Triton Inference Server to version 25.07 or later is currently the most important action to avoid exploitation of these vulnerabilities.
📉 Although no actual exploitations have been recorded so far, the widespread use of Triton makes the vulnerabilities a significant threat if not addressed promptly.
---
🛡️ Security vulnerabilities threaten modern technologies: The year 2025 witnessed a series of attacks in the fields of artificial intelligence and cryptocurrencies, with crypto losses exceeding $3.1 billion in just the first half of the year, due to vulnerabilities in smart contracts and mismanagement of privileges.
📌 Are you using any Nvidia AI solutions? Update immediately or contact your IT department.
#Artificial_Intelligence #Cyber_Security #Nvidia #AI_Security #Triton #Important_Updates #CryptoSecurity #CVE2025