OpenAI Codex Upgrade: AI Coding Tool Gains Desktop Control to Challenge Anthropic

OpenAI has stepped up its competition with Anthropic by rolling out a major upgrade to its AI coding assistant, Codex. The update signals a growing “AI coding war” between the two tech rivals as both companies race to dominate the market for automated development tools used by businesses and developers.

The most significant change is Codex’s new ability to operate directly on a user’s desktop in the background. This allows the AI to open apps, move the cursor, click, and type while the user continues working normally. According to OpenAI, this makes it possible to run multiple AI agents at once, helping with tasks like frontend testing, app development, and other workflows that don’t require constant user supervision.

OpenAI has also introduced an in-app browser feature, enabling Codex to interact with web-based tools and applications more directly. The company says this will be especially useful for game development and frontend design, with plans to eventually expand the feature so Codex can fully control browser activity beyond local environments.

Additional upgrades include a “memory” function that allows Codex to remember previous sessions and better understand a user’s working style. The tool also gains image-generation capabilities for creating mockups, presentations, and product visuals. On top of that, OpenAI has added 111 plug-in integrations with platforms like GitLab Issues and CodeRabbit, allowing Codex to automate tasks across different workplace tools.

The update highlights OpenAI’s push to strengthen its position in enterprise AI, where it is increasingly competing with Anthropic’s Claude Code. As both companies expand their agent-style tools, the rivalry is shaping up to redefine how developers and businesses use AI for everyday work, blending coding, automation, and workplace productivity into a single system.

source: techcrunch