ChatGPT Atlas, the new browser produced by OpenAI, enters headlong into a competition between tech giants that could be shaping the way in which we will interact with the Web in the future and above all with the information it contains: it is a real “browser war”, and it is not the first time this has happened. After Internet Explorer defeated Netscape Navigator in the 1990s (during what we could call the “first browser war”) and Google Chrome, starting from 2008, took the lead from Firefox and Internet Explorer itself (thus winning the “second browser war”), the ecosystem of software for browsing the Web seemed to have stabilized. Since then, Chrome has remained in the lead, but there are many signs that would indicate that a new digital conflict has already begun: a competition that some have already dubbed the “third browser war”. A war that is no longer played on the ground of speed or compatibility of software with various operating systems, but on the integration of artificial intelligence with the world of online search.
New (and also old) players in the tech world aim to change the very concept of the browser which, from a simple window for accessing online content, aspires to become a truly autonomous assistant, capable of responding, summarizing and, increasingly often, even carrying out actions autonomously. Perplexity, The Browser Company and OpenAI, but also already well-known players in the world of web search, such as Opera. Each of these, with its own “AI browser” equipped with intelligent agents capable of navigating and interacting with the Web in place of the user, tries to carve out its own space. And Google Chrome, despite maintaining the most important share of the current market shares, finds itself in a fragile position: on the one hand it has to face action from the US government’s Antitrust Authority, which accuses Google of a monopoly in search, on the other it suffers technological pressure from competitors who intend to rethink the very concept of web browsing and searching from scratch (and in response, Google has integrated AI Mode into its browser).
It’s still too early to say who will come out on top at the end of this third browser war. However, one thing is certain: in the end, online research will no longer be the same as before.
The protagonists of the third AI browser war: from Comet to ChatGPT Atlas
The companies leading this turning point have understood a lesson that Google had already learned twenty years ago: whoever controls the browser also controls search. So let’s see who the new protagonists of online search are.
Comet by Perplexity
Perplexity, the US startup that has become known for its chatbot-based response engine, offers Comet, an “AI-first” browser that combines traditional navigation with an intelligent agent capable of reading, understanding and summarizing the pages visited. The built-in assistant can summarize long articles, extract data, write emails or create calendar events without switching tabs. All of Perplexity’s search functions are incorporated directly into the browser: no links or results pages are needed, because the answers are shown in real time, like in a chat. Comet also supports voice and contextual commands and offers an “automatic mode” where the agent can navigate multiple sites to complete a task.
Initially available to subscribers to the Perplexity Max plan, from October 2025 it is available for free on Windows and macOS.
Dia by The Browser Company
The Browser Company, already known for the Arc browser, has developed Dia, a new navigation interface that puts artificial intelligence at the center of the experience. It can summarize documents, compare products, analyze PDFs or spreadsheets, and even suggest actions to take based on context (such as saving an event, generating notes or translating text). The History feature allows Dia to “remember” searches and sites visited in the last 7 days to offer personalized responses over time, a similar approach to “memory” more recent language models.
From October 2025 it is available for free to all users on macOS.
Neon by Opera
Opera, a veteran of the early browser wars, returns to the scene with Neon, a project that combines its classic Chromium engine with a human-like “contextual awareness” system. Neon is designed to understand what the user is trying to do, not just which pages they visit. Its AI engine interprets intentions and implements them via cloud agents connected to the user’s account. It can generate code snippets, work reports, articles, infographics and much more. Neon is also among the first browsers to offer offline local agents, capable of operating without a connection, thanks to lightweight AI models installed on the device.
To use the browser you must sign up to a waiting list. Furthermore, it is not free: it costs $19.99/month. It is available for Windows and macOS.
ChatGPT Atlas by OpenAI
OpenAI, the company that started the chatbot revolution with ChatGPT, also recently entered this new competition with ChatGPT Atlas, a browser that places the language model at the center of the browsing experience. Atlas allows you to search, visit and interact with the Web directly within ChatGPT, without going through lists of links: you can ask to “find sources on solar energy in Europe” and receive a complete summary complete with links to click on to learn more about the topic. Its “Agent mode” can also perform actions: fill out forms, manage reservations, etc. And thanks to the browser memories, Atlas is able to fully personalize the online browsing experience for each user.
It’s already available for free on macOS and is coming to Windows, Android and iOS.
The privacy issue
Behind the fierce technological competition that all the AI browsers mentioned in the article have in common lies a very important and delicate issue at stake: user privacy. Emails, messages, calendars and documents in general are increasingly interconnected, and new AI browsers promise to coordinate them automatically. However, this brings important new challenges for privacy. Intelligent agents have access to a much greater amount of personal data than traditional browsers and can infer intentions, habits, moods. Will these new research protagonists be able to do better than their predecessors? Difficult to predict at the moment. The companies that operate in this sector are certainly taking on a huge responsibility, as stated by a Fortunes the prof. George Chalhoub, ofUCL Interaction Centrewho bluntly said:
(Using browser AI, Ed.) is definitely a privacy risk, not because AI is inherently bad, but because it concentrates more context and intent in one place. So companies really need to assume their responsibilities in this regard.
