Google has launched a new program offering you the opportunity to earn 25 lakh rupees for just one task..

Google has launched its new AI Bug Bounty Program for cybersecurity researchers. This program is an extension of the company's previous Vulnerability Reward Program (VRP), but now specifically focuses on AI-related security flaws (AI vulnerabilities). The company states that the initiative aims to detect errors in which an AI system automatically takes an incorrect or unwanted action, such as unlocking a device, leaking data, or sending information to someone else's account.

According to Google, an "AI bug" is a weakness that can adversely affect a large language model (LLM) or generative AI system. For example, if an AI unlocks a Google Home door using AI, or if an AI steals email information and sends it to someone else, it would be considered a bug.

Google clarified that if the AI ​​simply creates misinformation or hate speech, it would not be considered a bug. Feedback for such cases would be provided within the product. This would be considered an AI flaw.

Rewards
Discovering serious AI bugs in Google Search, Gemini apps, Gmail, and Drive will earn a reward of $20,000 to $30,000. Finding bugs in smaller tools like NotebookLM or Jules Assistant will result in slightly lower rewards.

CodeMender - Bug Fixing with AI
Google has also launched CodeMender. This AI tool helps find and fix bugs in open-source software. So far, the tool has provided 72 verified fixes.

Why is it Important
Google says that as its products become more AI-based, their security should also strengthen. Since 2022, researchers have earned over $430,000 by finding AI-related bugs. Overall, this program is a good opportunity for those who want to work in AI security and make a name for themselves on a major platform like Google.

Disclaimer: This content has been sourced and edited from News 18 hindi. While we have made modifications for clarity and presentation, the original content belongs to its respective authors and website. We do not claim ownership of the content.