Google Will Pay Up to ₹26 Lakh to Anyone Who Can Find Serious Security Flaws in Its AI Systems

If you’re good at spotting flaws in tech, Google just opened a new way to earn some serious money. The company has launched a bug bounty program focused on artificial intelligence, with rewards that go as high as $30,000 — roughly ₹26 lakh.

The idea is pretty simple. If you can find a real security issue in one of Google’s AI systems — something that could cause harm or make it do something it shouldn’t — you get paid. This isn’t about making Gemini say something weird or inaccurate. It’s about stopping situations where an attacker could trick an AI model into leaking private data or controlling connected devices.

What Counts as a Real Bug

AI bugs work differently from normal software glitches. You’re not looking for broken code but for ways someone could manipulate how an AI behaves. Google’s calling these “rogue actions.”

For example, imagine an attacker crafting a sneaky command that makes an AI assistant summarize someone’s emails and send them to another account. Or maybe a prompt that convinces a smart home system to unlock a door. That’s the kind of thing Google wants to catch early.

But not every problem counts. If Gemini just hallucinates facts or says something inappropriate, that’s not a security issue. Those reports go through regular feedback tools instead, where Google’s safety teams use them to improve the model over time.

There’s Also an AI That Fixes Bugs

Alongside the bounty program, Google also announced something called CodeMender. It’s an AI agent that actually helps find and fix vulnerabilities in code. According to Google, CodeMender has already helped patch over 70 security issues in open-source projects after humans checked its work.

So in a way, Google’s using AI to help secure AI — and that’s kind of the point.

How the Rewards Work

The biggest rewards, up to $30,000, are for major issues found in flagship products like Gmail, Drive, Gemini Apps, and Search. Smaller payouts go to people who find less severe problems or bugs in other tools such as NotebookLM or Jules.

Over the last couple of years, researchers have already earned over $400,000 from Google’s AI-related reports. This new program just makes things more official — and a lot more competitive.

Why It Matters

AI systems are becoming a huge part of our daily lives. They’re in phones, laptops, home devices, and even the tools we use at work. The more capable these systems get, the more creative attackers can be. Google’s basically saying, “If you can break it before the bad guys do, we’ll pay you for it.”

That’s a pretty good deal for anyone with the right skills — and maybe a bit of curiosity.