Nvidia recently addressed two critical security vulnerabilities in Riva, an AI-powered speech and translation service, which could have allowed hackers to exploit AI systems.
Riva is a GPU-accelerated multilingual AI service designed for real-time speech-to-text, text-to-speech, and language translation. Nvidia disclosed the security risks in a March 10 advisory, detailing two access control flaws that left Riva instances vulnerable to cyberattacks.
Nvidia identified two security flaws in Riva:
- CVE-2025-23242: A high-severity vulnerability that enables privilege escalation, data tampering, denial of service (DoS), and information disclosure.
- CVE-2025-23243: A medium-severity vulnerability that allows data tampering and DoS attacks.
These vulnerabilities affect Riva versions 2.18 and earlier running on Linux. Nvidia has released a security patch in version 2.19.0 to mitigate these risks.
Researchers Discovered Security Vulnerabilities Facing Riva Instances
Security researchers from Trend Micro discovered these vulnerabilities and reported them to Nvidia in November 2024. The Trend Micro Zero Day Initiative later published detailed advisories on both flaws, emphasizing that attackers can exploit them without authentication.
Trend Micro security researcher Alfredo Oliveira explained that while Riva instances are not intended to be publicly accessible, their investigation revealed several misconfigured systems exposed to the internet.
“The default cloud installation creates a network rule exposing the service to 0.0.0.0/0, meaning it’s accessible to the entire internet,” Oliveira stated. This misconfiguration significantly increases the risk of cybercriminals exploiting unsecured Riva instances.
Oliveira also highlighted the financial consequences of these vulnerabilities. Since Riva powers AI-driven speech and translation services, running these systems requires substantial licensing fees and infrastructure costs. Attackers who gain unauthorized access to these services could cause significant financial damage to organizations relying on Riva for AI-powered applications.
Strengthening Security for AI Services
This incident serves as a critical reminder that AI-powered platforms must prioritize strong security measures, including proper access controls and configuration management. Organizations using Riva should immediately update to version 2.19.0 and review cloud access permissions to prevent exposure to unauthorized users.
