On Tuesday, OpenAI launched its first artificial intelligence-powered web browser. Media reports indicate this marks the ChatGPT developer’s entry into a new front of competition with Alphabet-owned Google, as a growing number of internet users rely on AI for information.
OpenAI has named this browser ChatGPT Atlas. It is designed to provide users with a more personalized web browsing experience while also performing tasks on their behalf, such as booking flights or editing documents. The company stated that the browser is built around ChatGPT; whenever a user visits a webpage, an “Ask ChatGPT” option appears on the page. Clicking it opens a sidebar for interacting with ChatGPT. For instance, when opening a movie review, a user can ask ChatGPT to summarize it, or after finding a recipe, have ChatGPT help order the necessary ingredients online.
OpenAI CEO Sam Altman stated during a live event on Tuesday: “This is a browser powered by AI, with ChatGPT at its core. We believe AI provides a once-in-a-decade opportunity to rethink what a browser can be. This is an all-around excellent browser—smooth, fast, and very comfortable to use.” OpenAI stated that Atlas is now being released globally for macOS, with versions for Windows, iOS, and Android to follow. Currently, only paying users (ChatGPT Plus and Pro subscribers) will have access to the more advanced AI agent features.
OpenAI mentioned that it currently has over 800 million users on ChatGPT, but many of them use the free version. OpenAI is still operating at a loss and is seeking new ways to achieve profitability. Media reports suggest that by entering the web search portal with its own browser, OpenAI, the world’s highest-valued startup, could potentially attract more web traffic and thereby generate revenue from digital advertising.
The latest news also note that OpenAI thereby joins a group of companies leveraging AI to reimagine browser products, a category dating back to the 1990s. AI startup Perplexity AI recently launched the Comet browser, designed to function as a digital assistant capable of automatically completing tasks for users. Both Opera and The Browser Company are also adding more AI features to their respective browsers.
ChatGPT Integrates into Sidebar, Features Built-in AI Agent
Ben Goodger, the engineering lead at OpenAI responsible for the Atlas project, stated during Tuesday’s live event that ChatGPT is the core of the company’s first browser. Within ChatGPT Atlas, users can converse with search results, similar to the experience in Perplexity or Google’s AI mode.
A killer feature of other AI browsers is the built-in side panel—where a chatbot automatically captures contextual information from the user’s screen. This might sound minor, but many users frequently copy-paste text, drag files, or links into ChatGPT to provide context. This sidebar feature eliminates these cumbersome steps, making the experience much smoother. Adam Fry, Head of Product at OpenAI, mentioned during the live event that ChatGPT Atlas is also equipped with this “sidecar” functionality. Additionally, Atlas includes a “Browsing History” feature, meaning ChatGPT can record the websites users visit and their actions, providing more personalized responses based on this data.
In Atlas, whenever you click a link from search results, the default view is a split-screen interface: the webpage on the left and the ChatGPT conversation history on the right. This design aims to keep the AI “assistant” readily available. Users can also turn off the split-screen mode. During the live demo, employees showcased the browser’s webpage summarization feature and how users can select a sentence in an email and click a button to have ChatGPT instantly refine it—a feature called “cursor chat.”
Another common characteristic of AI browsers is a built-in AI agent, intended to help users automatically complete tasks on the web. Based on media tests, early versions of web AI agents still need improvement. While Perplexity’s Comet and OpenAI’s ChatGPT agent perform well on simple tasks, they still struggle with the complex tasks users truly want to delegate to AI. As expected, OpenAI’s browser also includes web agent capabilities. Users can activate “Agent mode” to let ChatGPT handle minor tasks within the browser. The company stated that this feature will initially be available only to ChatGPT Plus, Pro, and Business tier subscribers at launch.
Fry stated, “In Atlas, ChatGPT can now take actions for you… It can help you book a restaurant or a flight, or even directly edit the document you are working on.” At OpenAI’s DevDay conference, ChatGPT lead Nick Turley told media in an interview that he was inspired by how browsers redefined the concept of the operating system. Turley pointed out that browsers revolutionized how people work online, and he sees ChatGPT as a similarly revolutionary phenomenon.
