F5 Networks Indonesia Unveils Two New Tools to Mitigate AI Risks

F5 Networks, a US-based application connectivity (API) security company, has introduced two new approaches to secure the use of artificial intelligence (AI) within companies in Indonesia.

The two service-based approaches are AI Guardrails and AI Red Team.

Country Head of Solutions Engineering at F5 Networks Indonesia, Danang Wijanarko, stated that these solutions are designed to minimise the risks arising from the use of AI in a company’s API and cloud ecosystem.

“These risks can arise because AI is non-deterministic and lacks a fixed pattern. It is generative in nature. Whatever input is provided, it will respond and accept it, regardless of whether it is harmful or not. This is what needs to be regulated and monitored,” Danang added.

In simple terms, AI Guardrails can be understood as a “safety fence” installed before users connect to the AI system.

Danang mentioned that many companies in Indonesia currently use AI models that run on the public internet (cloud).

To enable AI to answer questions according to business needs, the system typically combines user questions with internal company data. This method is known as Retrieval Augmented Generation (RAG).

The problem is that this process carries the risk of sensitive data stored in the company’s database not being filtered and being sent outside the company and the AI cloud system.

“This is where the term ‘oversharing’ emerges, where company data that should not be stored is leaked and received and processed by the AI. Usually, this oversharing to AI is not realised by a company,” Danang explained.

More specifically, AI Guardrails has the following functions:

In other words, AI Guardrails can ensure that companies can continue to leverage AI without losing control over their data.