Google’s AI bounty program pays bug hunters up to $30K

Google has launched a new reward program that pays bug hunters up to $30,000 for finding bugs in its AI products. The program focuses on finding issues that allow AI systems to cause harm or take advantage of security loopholes, such as modifying someone's account or data, or generating harmful content like hate speech or copyright-infringing material. The program clarifies what constitutes an AI bug, with rogue actions using large language models or generative AI systems being the top priority. Bug hunters have already earned over $430,000 during the two years since Google started inviting AI researchers to find potential avenues of AI abuse. In addition to the new AI reward program, Google also announced an AI agent called CodeMender that patches vulnerable code in open-source projects, with 72 security fixes already implemented. The $20,000 prize is awarded for finding rogue actions on Google's flagship products, with multipliers and bonuses available for high-quality reports and novel findings. The reward amount is lower for bugs found on other Google products or for less severe issues.