Google has launched CodeMender, an AI-powered tool that autonomously detects and repairs software vulnerabilities, revolutionizing cybersecurity maintenance. Fuelled by the Gemini Deep Think models, CodeMender generates and applies patches for critical flaws, enabling developers to prioritize building secure software instead of manually fixing issues. Over six months, it has contributed 72 security fixes to open-source projects, utilizing a combination of static and dynamic analysis, fuzzing, and differential testing to accurately trace and validate vulnerabilities. CodeMender operates with multi-agent collaboration to enhance decision-making, and each fix is rigorously validated before human review to ensure compliance with coding standards. Its proactive approach addresses both new vulnerabilities and eliminates entire classes of security flaws. Google asserts that CodeMender’s innovative capabilities signify a significant advancement in AI-driven cybersecurity, showcasing how artificial intelligence can fundamentally enhance software protection and maintenance. Interested in AI, tech, and digital diplomacy? Consult our Diplo chatbot for more insights!
Source link
