OpenAI announced on Tuesday that it is launching a new web browser named Atlas, which integrates directly with ChatGPT. The browser features a sidebar that allows users to ask ChatGPT questions about the web pages they’re viewing. It also includes an AI agent capable of navigating sites and completing tasks on behalf of the user.
“We believe AI offers a once-in-a-decade opportunity to completely rethink what a browser can be,” said OpenAI CEO Sam Altman during the Atlas launch livestream. “Tabs were revolutionary, but browser innovation has largely stalled since then.”
The debut of Atlas comes as Silicon Valley intensifies its race to harness generative AI in transforming how people experience the internet. Google, for instance, has rolled out a range of AI-powered features for its popular Chrome browser, including a ‘sparkle’ button that activates its Gemini chatbot. Despite the growing competition, Chrome continues to be the most widely used browser worldwide.
OpenAI announced that the Atlas browser is now available worldwide for ChatGPT users on macOS starting today. Versions for Windows and mobile devices are currently in development. While Atlas is free to use, its AI agent capabilities are exclusive to subscribers of ChatGPT Plus and ChatGPT Pro plans.
“We’ve introduced significant upgrades to ChatGPT’s search experience when accessed through Atlas,” said Ryan O’Rouke, OpenAI’s lead designer for the browser, during the livestream. For example, when a user searches for movie reviews in the Atlas search bar, they’ll first see a chatbot-style response rather than the traditional list of blue links typically shown by Google.
In addition to the AI-generated response, users can switch between tabs to view a curated set of website links, images, videos, or news articles related to their search. It’s somewhat the reverse of the Google Chrome experience — instead of starting with a list of links enhanced by AI, Atlas puts the AI chatbot front and center, with traditional search results like links and media displayed as secondary options.
Another feature showcased during the livestream is Atlas’ “browser memories.” This optional capability builds on ChatGPT’s existing memory feature, which stores contextual details based on a user’s past interactions. With browser memories enabled, Atlas can recall previous searches and use that information to suggest topics of interest or recommend actions — such as automating repetitive online routines or revisiting a website previously accessed that may be relevant to a current project.
For several years, both tech giants and smaller startups have been experimenting with integrating AI into web browsers. Microsoft was an early mover, embedding its AI tool, then called Bing, into the Edge browser as a sidebar. Since then, browser-focused companies like Opera and Brave have continued exploring different AI integrations. Another notable entry in the AI browser space is Perplexity’s Comet, launched this year and also free to use.
Although Atlas marks OpenAI’s first standalone browser, it isn’t the company’s first attempt at merging generative AI with web browsing. Earlier this year, OpenAI introduced an agent tool within ChatGPT that can carry out tasks on a user’s behalf — from comparing products to suggesting which options might best suit a person’s preferences and lifestyle. When WIRED tested the bot’s ability to navigate the web, conduct research, and even play chess, the results were sometimes slow and occasionally inaccurate.
