OpenAI announced Tuesday the launch of its AI-powered browser, ChatGPT Atlas, a major step in the company’s quest to unseat Google as the main way people find information online.
The company says Atlas will first roll out on MacOS, with support for Windows, iOS and Android coming soon. OpenAI says the product will be available to all free users at launch.
Browsers have quickly become the AI industry’s next battleground. While Google Chrome has long dominated the space, there’s a sense that AI chatbots and agents are fundamentally changing how people get work done online. A handful of startups have tried to capture this by launching AI-powered browsers of their own, such as Perplexity’s Comet and The Browser Company’s Dia. Google and Microsoft have also tried to update Chrome and Edge, respectively, with AI-powered features to make their legacy products stand out.
OpenAI’s Engineering Lead for Atlas, Ben Goodger, said in a livestream Tuesday that ChatGPT is core to the company’s first browser. Users in ChatGPT Atlas can chat with their search results, much like in Perplexity or in Google’s AI Mode.
The killer feature for other AI-powered browsers has been the built-in chatbot that sits in a side panel and automatically has context for whatever’s on your screen. It may sound minor, but many users spend all day copying and pasting text or dragging files and links into ChatGPT, just to provide context. The sidecar feature removes that friction, and makes for a smoother user experience.
OpenAI’s Product Lead Adam Fry said during the livestream that ChatGPT Atlas will have the sidecar feature, too. Further, ChatGPT Atlas has “browser history,” meaning that ChatGPT can now log the websites you visit and what you do on them, and use that information to make its answers more personalized.
AI-powered browsers also commonly feature an AI agent that aim to automate web-based tasks on behalf of users. In TechCrunch’s testing, we’ve found the early versions of web-browsing AI agents leave something to be desired. While Perplexity’s Comet and OpenAI’s ChatGPT agent work well for simple tasks, they struggle to reliably automate the more cumbersome problems users might want to offload to an AI system.
Techcrunch event
San Francisco
|
October 27-29, 2025
Sure enough, OpenAI’s browser has a web-browsing agent too. By using “agent mode,” users can ask ChatGPT to complete small tasks in the browser on their behalf. The company says agent mode is only available to ChatGPT users on the Plus, Pro, and Business tier at launch.
In an interview at OpenAI’s DevDay conference, Head of ChatGPT Nick Turley told TechCrunch that he’s inspired by the way browsers have redefined what an operating system can look like. Turley noted that browsers have revolutionized the way people get work done online, and he thinks ChatGPT is a similar phenomenon.
Whether OpenAI’s browser can put a dent in Google Chrome, which has more than three billion users around the globe, remains to be seen. AI browsers are quite buzzy in Silicon Valley today, but their impact in the broader world is limited today.
