Zscaler's Jay Choudhury on India’s AI Boom & The Rising Threat of Agentic AI

So first of all, thank you for this opportunity. Our report comes from the data we have. We see all communication of employees of enterprises engaging with internet with AI and whatnot. So from enterprise employees' point of view, India is actually very active in the business world using AI. That was number one. I was surprised to see that. Number two, people are beginning to use AI, especially agentic AI, to attack things. Just like any technology, AI is very powerful but AI is very dangerous. You can use a kitchen knife for good and the bad stuff. What AI is doing is automating and making it simpler to attack things. For example, with AI, you can easily find all these different firewalls of a given company that may have security vulnerabilities. You can attack them and get inside. You can use AI to create sophisticated phishing emails where employees fall for it and make things happen. But it's not all doom and gloom. We believe you can fight AI with AI. Zscaler is using AI technology to actually do things that could not be easily done before. I'll give you a couple of examples. Security is never 100% foolproof. It's a race with bad guys. You build a defense, the bad guys find a way around it. AI can be used to find anomalous behavior. For example, a user generally goes and accesses application A, B or C. If that identity got stolen, now a bad guy is trying to go under the other person's name. Now he's looking for all kinds of applications that had never been discovered or looked at before. AI can pick up that behavior and tell us, alert us and say something fishy is going on. This can be done very quickly with AI and we are doing that today for our customers. When data is leaked out there, that's a big problem. AI can be used to leak data. Employees are submitting sometime confidential information and say AI check my source code or check this information or rephrase a confidential email I wrote. Now AI will rephrase it. But public AI such as ChatGPT will have that data out there. So essentially the data is leaked out there. Now we can easily use AI to classify the type of documents being sent and put a policy that says this kind of data should not be submitted out there. So it is powerful. The main message I leave with for enterprises is that hackers have no inertia. They move fast. Large companies are slow, have inertia. They should embrace these defenses at a faster pace. Everything in the world is moving at a faster pace and cyber must move faster.