Cover An illustration photo shows ChatGPT Atlas logo in a smartphone in China (Photo: CFOTO/Future Publishing via Getty Images)
An illustration photo shows ChatGPT Atlas logo in a smartphone in China (Photo: CFOTO/Future Publishing via Getty Images)
1 / 1
OpenAI enters the browser wars with revolutionary AI integration through ChatGPT Atlas
On October 21, 2025, OpenAI launched ChatGPT Atlas, marking the company’s bold entry into the web browser market and setting up a direct confrontation with Google Chrome’s dominance. The announcement sent ripples through the tech industry, with Alphabet shares dropping over 2 per cent as investors weighed the implications of this new competitor.
The move marks a pivotal moment in the AI race, as OpenAI transitions from being primarily a chatbot provider to challenging one of Google’s most valuable assets. With over three billion users relying on Chrome daily, OpenAI’s entry into this space signals its ambition to become an indispensable web platform.
In case you missed it: Menopause isn’t just a transition—it’s a hormone deficiency with health risks
What is ChatGPT Atlas?
Above ChatGPT Atlas browser is based on Chromium, the same open-source engine that powers Google Chrome and Microsoft Edge (Photo: Pixabay)
ChatGPT Atlas browser is based on Chromium, the same open-source engine that powers Google Chrome and Microsoft Edge (Photo: Pixabay)
1 / 1
ChatGPT Atlas is an artificial intelligence-powered web browser built with ChatGPT integration at its core. Unlike traditional browsers that have added AI features as an afterthought, Atlas was designed from the ground up around the concept of conversational browsing. As OpenAI CEO Sam Altman explained during the launch livestream, “We think that AI represents a rare once-a-decade opportunity to rethink what a browser can be about and how to use one.”
The browser is based on Chromium, the same open-source engine that powers Google Chrome and Microsoft Edge, but it differentiates itself through deep integration with ChatGPT rather than competing on infrastructure.
Read more: 5 times ‘Black Mirror’ predicted the future of AI
ChatGPT Atlas’s key features and capabilities
AI represents a rare once-a-decade opportunity to rethink what a browser can be about and how to use one
The centrepiece of ChatGPT Atlas is its “Ask ChatGPT” sidebar, which appears on every webpage and automatically understands the context of what you’re viewing. Gone are the days of constantly copying and pasting content between your browser and ChatGPT. The AI can instantly summarise articles, compare products across shopping sites, analyse data and code, help edit documents and emails, or provide explanations of complex content without leaving the page. This seamless integration means the assistant is always aware of what’s on your screen, ready to help with whatever task you’re working on.
But ChatGPT Atlas goes far beyond simple assistance. The browser introduces browser memories, allowing ChatGPT to remember sites you visit and recall that information later. Imagine asking, “Find all the job postings I looked at last week and create a summary of industry trends for my interviews”—and getting the results in seconds.
However, users maintain complete control over these memories, with options to view, archive, or delete them at any time.
The most ambitious feature is Agent Mode, available to Plus, Pro, and Business subscribers. Agent Mode works by allowing ChatGPT to actually interact with websites on your behalf: it can click buttons, fill in forms, navigate between pages and complete multi-step processes. Users can intervene at any point through “take control” and “stop buttons”—it’s a bit similar to screen sharing with your IT team.
Read more: AI decoded: what those buzzwords really mean
It all sounds very good—but does it really work?
Above ChatGPT Atlas seems to excel at simple tasks, but struggles with complex, multi-step operations involving multiple tabs and browser history (Picture: Pixabay)
ChatGPT Atlas seems to excel at simple tasks, but struggles with complex, multi-step operations involving multiple tabs and browser history (Picture: Pixabay)
1 / 1
Early reviews show promise but reveal rough edges. While the sidebar excels at simple tasks, browser memories can be inconsistent. Some testers found that Atlas failed to recall recent, frequent searches from imported browsing history.
The Agent Mode also seems to struggle with complex, multi-step operations. For example, it might successfully book a simple restaurant reservation, but struggle with something like “research flights to three different cities, compare hotel options in each, and create a spreadsheet comparing total costs.”
Still, Atlas represents OpenAI’s boldest move yet to become an essential computing platform rather than just an AI chatbot company. Whether it can chip away at Chrome’s three billion users remains the ultimate test.
Topics
Suchetana Mukhopadhyay is the Digital Content Manager for Tatler Hong Kong. In this role, she leads all digital editorial and branded content on Tatler Hong Kong’s website, from brainstorming story ideas with the writers to editing and publishing the articles, and from managing the overall content flow to driving search engine optimisation. She also leads the beauty content at Tatler Hong Kong and across the region, and is always looking to champion diversity through her articles. She was previously with Cedar Hong Kong and Gafencu, and freelanced for the South China Morning Post, Campaign Asia-Pacific, CNN and more. Contact her here.
