OpenAI Launches Atlas: A ChatGPT-Powered Browser to Challenge Chrome

OpenAI launched Atlas, a ChatGPT-powered browser for Mac, featuring AI-driven navigation and “agent mode.” While it aims to simplify online tasks, experts warn of privacy and autonomy concerns. It enters a Chrome-dominated market, amid rising AI use despite accuracy issues.

The US artificial intelligence company OpenAI announced on Tuesday that it is launching its own web browser, named Atlas, with the goal of rivaling Google’s widely popular Chrome browser. This marks a significant move for the California-based company, which has become globally known for its AI chatbot, ChatGPT, and signals its ambition to transform the way people interact with and navigate the internet.

Atlas will be powered directly by ChatGPT, OpenAI’s advanced conversational AI system, and is designed to provide a more integrated and intelligent browsing experience. OpenAI’s CEO, Sam Altman, highlighted the novelty of the project in a video presentation broadcast on Tuesday, saying, "Tabs were great, but we haven't seen a lot of browser innovation since then." He emphasized that Atlas represents a “rare, once-a-decade opportunity to rethink what a browser can be about and how to use one," suggesting that the company sees this as a chance to fundamentally reshape the concept of web browsing.

One of the most notable differences in Atlas is the potential replacement of the traditional URL search bar with an AI chatbot interface. Rather than typing queries into a search bar and clicking through multiple links, users could interact with the AI to get precise answers, summaries, and recommendations directly within the browser. This functionality is intended to reduce the need for copying and pasting information from one page to another, creating a smoother and more efficient online experience.

Initially, Atlas will be available only for Apple’s Mac computers, according to OpenAI. The browser is being marketed as a tool that can help users complete tasks seamlessly on a single platform, minimizing the interruptions typically associated with traditional browsing. A key feature is what OpenAI calls "agent mode," which allows the browser to navigate the internet automatically on behalf of the user. By leveraging a person’s browsing history and predicting the type of content they are likely looking for, Atlas can proactively gather information and anticipate user needs. Altman described this as "using the internet for you," suggesting that the browser would become an intelligent assistant capable of performing complex tasks autonomously.

However, some experts have expressed caution about the implications of this level of automation. Paddy Harrington, an analyst with London-based market research group Forrester, warned that Atlas could, in effect, "take personality away from you." He explained that the browser would create a highly personalized profile based on extensive data collection, raising questions about authenticity and user autonomy. "Your profile will be personally attuned to you based on all the information sucked up about you," Harrington told the Associated Press. "OK, scary. But is it really you, really what you're thinking, or what that engine decides it's going to do? And will it add in preferred solutions [to users' queries] based on ads?" Despite these concerns, he acknowledged that the concept represents a bold technological step, though he cautioned that it would face significant challenges in competing with Google, a company with overwhelming market dominance.

Google Chrome, which launched in 2008, has grown to roughly 3 billion users worldwide, outpacing competitors such as Microsoft’s Internet Explorer and later the Edge browser. Chrome’s dominance has created a formidable barrier to entry for any new browser. Nevertheless, the growing prevalence of AI chatbots like ChatGPT, which can quickly summarize information from the web, has shifted some user behavior away from traditional link-clicking. People increasingly rely on AI tools to provide concise, synthesized information rather than navigating multiple pages to find answers.

OpenAI has stated that ChatGPT already boasts more than 800 million users. A survey commissioned by the AP this year revealed that approximately 60% of Americans, and 74% of those under 30, use AI to seek information at least some of the time. Major browsers, including Chrome, have also started integrating AI-generated summaries into their search results, typically displayed at the top of the page above the first link. However, these summaries are not without controversy, as concerns have emerged about their reliability and accuracy.

On Wednesday, research published by the European Broadcasting Union (EBU) and the BBC found that 45% of AI responses studied contained at least one significant error, while 81% had some form of problem. The study evaluated AI assistants in 14 languages, assessing their accuracy, sourcing, and ability to distinguish opinion from fact. This included tools such as ChatGPT, Copilot, Gemini, and Perplexity. In one notable instance, ChatGPT incorrectly reported that the current Pope was Pope Francis, several months after his death, illustrating how confidently AI can deliver inaccurate information. Such mistakes have the potential to undermine public trust and reliability in AI systems, with consequences that experts warn could be harmful.

Jean Philip De Tender, Media Director at the EBU, emphasized the societal implications of AI misinformation, stating, "When people don't know what to trust, they end up trusting nothing at all, and that can deter democratic participation." His comments highlight the broader stakes involved in the increasing reliance on AI-driven information, particularly as tools like Atlas seek to redefine how users engage with the web.