A10 Networks is a US-based technology security company that typically provides security solutions for critical IT infrastructure in offices, cloud data centers, and AI data centers.
We’ve previously read about incidents of how custom or commercial language models like ChatGPT, Gemini, and others can be vulnerable to cyber threats, just like many, if not all, other digital and online services.
A10 Networks’ firewall application combines GPU processing power and security software that can be placed in front of APIs and AI and LLM service links to ensure that data sent to AI services is checked and filtered before being received and processed by the AI service.
A10 Networks says that their firewall application can be used for both machine learning and inference purposes by custom and commercial AI services developed by companies like OpenAI, Anthropic, Google, and others.
The application can detect AI threats such as prompt injection and confidential information leaks by inspecting commands and responses at the command level, and enforce the security policies needed to address those threats.
It can also detect if there are network or processing capacity issues between the AI data center and the user. The system built into the A10 Networks firewall can give AI service owners several real options to increase service capacity before the issue becomes critical.
Over time, the service will also learn the usage trends of the company's AI systems. In addition to providing accurate estimates of when their services will be used most frequently, it will also inform the company's IT team about when it is best to perform maintenance and updates on their IT systems.
A10 Networks is demonstrating the capabilities of its AI firewall application at the Interop Tokyo conference, which is taking place throughout this week.