Олександр Кузьменко AI Eng 7 October 2025, 18:40

Google has created an AI agent called CodeMender that can automatically detect and fix errors in software code

Google researchers have introduced CodeMender, an artificial intelligence-based agent that helps improve code by patching critical bugs.

Leave a comment

Google has created an AI agent called CodeMender that can automatically detect and fix errors in software code

Google researchers have introduced CodeMender, an artificial intelligence-based agent that helps improve code by patching critical bugs.

This allows developers to focus on creating reliable software, rather than manually finding and fixing flaws, notes Digital Watch.

CodeMender, built on Gemini Deep Think models and operating autonomously, analyzes the underlying code and fixes bugs it finds. In the past six months, it has made 72 security fixes to open source projects, including those containing millions of lines of code.

«While our initial results with CodeMender are promising, we are taking a cautious approach with a focus on reliability. Currently, all patches generated by CodeMender are reviewed by human researchers before they are sent up the chain,» Google says.

The company believes that CodeMender’s dual approach (reactively fixing new flaws and proactively rewriting code to eliminate entire classes of vulnerabilities) is a significant step forward in the field of artificial intelligence-based cybersecurity.

Google previously announced updates to Jules Tools — a command line interface (CLI) and public API that allow it to be integrated directly into developer workspaces.