---
title: Google enhances vulnerability-fixing AI CodeMender using multiple AI models
url: https://www.dataloco.com/en/google-enhances-vulnerability-fixing-ai-codemender-using-multiple-ai-models
published: 2026-11-09T18:20:41+00:00
language: en
section: Security
source: https://japan.cnet.com/article/35254533/
organizations: google, codemender, gemini, argon
publisher: Dataloco
---

# Google enhances vulnerability-fixing AI CodeMender using multiple AI models

Google has strengthened its AI, CodeMender, which discovers and fixes software vulnerabilities. The company announced that users can now select a sandbox environment for validating vulnerabilities and choose the scope of files to investigate, as detailed in an official blog on November 4. CodeMender is currently available in a preview version.

The validation environment offers options to use a sandbox that limits program operations with OS functions or to build a virtual machine that simulates an independent computer. Executing vulnerability validations or reproducing attacks in an isolated environment reduces the risk in case issues arise during validation. Additionally, a system has been added that requires human confirmation for writing files or executing commands in the local environment.

Users can choose from three options for the investigation scope: specific folders, entire repositories, or only files that have changed since the last investigation. By narrowing down the target, such as examining only changed files, time and costs associated with the investigation can be reduced.

CodeMender utilizes multiple AI models, including Google's Gemini 4 Argon. It is said to identify vulnerabilities that had not been discovered previously and generate verified fixes before attackers can exploit them. Since the preview launch in July, Google has explained that users can reproduce attacks in isolated environments. The recent enhancements allow users to select the type of isolation environment and specify the investigation scope. However, developer confirmation and approval remain necessary before reflecting AI-generated fixes in the repository.

## This story in other languages

- [Türkçe](https://www.dataloco.com/tr/google-codemender-adli-yapay-zekayi-yeni-ozelliklerle-guncelledi)
- [Italiano](https://www.dataloco.com/it/google-potenzia-lai-per-la-correzione-delle-vulnerabilita-software)
- [Español](https://www.dataloco.com/es/google-fortalece-el-ai-de-correccion-de-vulnerabilidades-codemender)
- [Português (Brasil)](https://www.dataloco.com/pt-br/google-fortalece-ai-de-correcao-de-vulnerabilidades-codemender)
- [Nederlands](https://www.dataloco.com/nl/google-versterkt-ai-voor-kwetsbaarheidsdetectie-codemender)