OpenAI’s next move in the fight against Google is an artificial intelligence-based web browser. A tool called ChatGPT Atlas is available today. The company made the announcement on a livestream after teasing it earlier on Tuesday via a cryptic message video browser bookmarks on white screen.
ChatGPT Atlas is available “globally” on macOS starting today, with access for Windows, iOS and Android “coming soon,” according to the company. However, “Agent Mode” is only available to ChatGPT Plus and Pro users for now, OpenAI CEO Sam Altman said in a livestream.
“The way we hope people will use the Internet in the future … web browser chat might be a great analogy,” Altman said.
In addition to Altman, the livestream included OpenAI employees Will Ellsworth, who works on post-training research; Adam Fry, lead product at ChatGPT search engine; Ben Goodger, an employee who in previous positions helped develop Google Chrome and Mozilla Firefox browsers; Ryan O’Rouke, interface designer; Justin Rushing, who previously worked at Apple; and Pranav Vishnu.
Fry said one of the browser’s best features is memory, which makes it “more personalized and more helpful to you,” as well as Agent Mode, which means that “within Atlas, ChatGPT can now take actions for you… It can help you book reservations or flights, or even just edit the document you’re working on.” According to employees, users can view and manage browser “memories” in settings, as well as open incognito windows.
Browser Agent Mode clearly builds on OpenAI’s previous forays into agentic AI, such as the Operator tool, an early version of the tool that allows ChatGPT to exploit the computer on the user’s behalf, and ChatGPT Agent, the next iteration that was designed to be able to perform more elaborate tasks, make purchases, and more (though it wasn’t always successful in this area).
Whenever you click on a link in an Atlas search result, it defaults to a split screen with the web page and ChatGPT transcript, with the goal of continuously displaying the “companion,” workers say, though the user can turn off the split screen if they prefer. During the livestream, employees also demonstrated browser features for summarizing web pages, as well as selecting text from an email and clicking a button to have ChatGPT reorder the sentence on the fly – the latter feature called “cursor chat.”
“It’s just a great all-around browser – it’s smooth, it’s fast and it’s really nice to use,” Altman said.
The AI browser wars have been heating up for some time now – OpenAI announced a prototype of its search engine, named SearchGPTuntil July 2024. However, in recent months, AI-powered browsers seemed to be the latest trend in the AI hype.
This summer, Perplexity released the powerful Comet Browser, an AI-powered solution that aims to simplify the way people browse the web and perform tasks. Instead of a dehydrated list of Google search results, you get Perplexity’s “answer engine”, which offers several links to relevant websites and generates an answer to your query. It can also scan all your open tabs, summarize videos, organize your inbox, and even shop on Amazon.
In September, Google announced that it would embed its Gemini AI assistant more deeply into Chrome, and that in the coming months it planned to enable Gemini in Chrome to perform “boring tasks” on your behalf, such as grocery shopping, scheduling meetings, reserving books, and more – although Google declined to provide a release date.