Google's New AI Agent, Code Mender, Patches Open-Source Vulnerabilities
Google has introduced a new AI agent, Code Mender, designed to patch vulnerable code in open-source projects. The agent has already helped close 72 security loopholes, with human review ensuring accuracy.
Code Mender is part of Google's swarm intelligence initiative, crowdsourcing vulnerability discovery. The tech giant has paid over $430,000 to external developers in the past two years for finding AI function misuse.
Google's rewards program categorizes potential vulnerabilities and offers prizes ranging from $20,000 to a potential $30,000, depending on the severity and novelty of the findings. The company has not disclosed the name of the firm involved in Code Mender's development or the specific open-source projects it has secured.
Code Mender's launch demonstrates Google's commitment to improving AI security. The rewards program encourages external developers to contribute, fostering a collaborative approach to vulnerability discovery.
