AI Browser Failures: Why Most Projects Fell Apart

The emergence of AI browsers, once hailed as a "once-a-decade opportunity" by figures like Sam Altman, has largely resulted in failure and retraction. Despite significant investment and hype, most AI browsers, including OpenAI's ChatGPT Atlas and Microsoft's Copilot Mode in Edge, have been shut down or repurposed, failing to gain significant user adoption.

The Rise of AI Browsers

Between May and October 2025, five companies either launched or announced AI browsers, each promising an agent that could browse the web autonomously.

Key players and their offerings included:

• Dia by The Browser Company: Known for its popular Arc browser, Dia featured a search bar that doubled as a chatbot, a sidebar assistant capable of reading all open tabs, and "skills" – saved prompts for automation. The Browser Company raised $128 million and was valued at $550 million by March 2024.
• Opera Neon: A subscription-based browser ($19.99/month) with an agent called "Neon Do" designed to fill forms, compare products, and order groceries. It also included "Cards" for reusable prompts.
• Perplexity Comet: Launched by the AI search engine Perplexity.
• Microsoft Edge's Copilot Mode: Microsoft integrated Copilot into Edge, aiming for an anticipatory browsing experience.
• Google's Project Mariner: Unveiled at I/O 2025, this browser agent was part of a $250/month AI Ultra plan, specifically built for browser use.
• OpenAI's ChatGPT Atlas: Launched a few months after Google's announcement, Atlas integrated ChatGPT into every page, remembered visited sites, and performed tasks via an agent mode. Initially available only on Mac, its launch caused Alphabet's stock to drop 2%.

Why the Obsession with Browsers?

AI companies were drawn to browsers because they represent a central hub of daily digital activity. Capturing this layer offered lucrative opportunities for advertising, subscriptions, and, crucially, data collection. As Perplexity's CEO explained in April 2025, the goal was to gather data beyond the app to build better user profiles and potentially serve targeted ads. Browsers are a "goldmine" of user preferences, habits, and purchasing information.

The Downfall of AI Browsers

Despite the initial enthusiasm, most AI browsers failed to thrive:

• Microsoft Copilot Mode in Edge: Retired by May 2026, its browsing agent was limited to Microsoft 365 Premium subscribers in the US. Even Microsoft executives questioned its utility.
• Google's Project Mariner: Shut down by May 2026 due to being slow, expensive, and error-prone. Its technology was absorbed into other Google products.
• OpenAI's ChatGPT Atlas: OpenAI began warning staff against "side quests" by March 2026, viewing Atlas as one such distraction. It was announced in July that Atlas would cease operations on August 9, 2026, less than a year after its launch and without ever expanding beyond Mac. OpenAI subsequently integrated Atlas features into existing products like the ChatGPT desktop app and a Chrome extension. This wasn't OpenAI's first attempt; "Operator," launched in January 2025, met a similar fate, becoming a ChatGPT agent after seven months. Ironically, in May 2026, Atlas was named the "breakout winner" of the AI browser race by presenc.ai, with an estimated 10-15 million monthly users, only to be shut down months later.

Survivors and Pivots

A few AI browsers managed to survive, albeit with shifts in focus:

• Opera Neon: Still exists but has not been actively promoted by Opera, even in earnings calls discussing AI integration in their other browsers.
• Dia: The Browser Company initially tried to integrate AI features into its Arc browser, but this led to a lack of cohesion and user confusion. Arc was eventually discontinued, and users were pushed towards Dia. Atlassian acquired The Browser Company for $610 million, aiming to transform Dia into a "doer, not a browser," a tool capable of more than just web browsing. Dia had yet to release a Windows version by September 2026.
• Perplexity Comet: Launched in July 2025, it became free by October, arrived on iOS in March 2026, and continued receiving updates. It also won an appeal against Amazon in August. Perplexity's persistence stemmed from its inability to make its AI assistant the default on phones due to Google's existing agreements, making a browser an alternative route. Comet is built around Perplexity's AI search engine, giving existing users a reason to switch, though it defaults to Google for navigational searches on iOS. Comet boasts 10 million monthly active users, but this pales in comparison to Firefox (198 million MAU with 2.8% market share) and Chrome. Perplexity even attempted to buy Chrome from Google for $34.5 billion.

Reasons for Failure

Several factors contributed to the widespread failure of AI browsers:

1. User Inertia: Most users stick with the default browser on their devices. Chrome holds 69% of the global market, and Safari nearly 16%. Even OpenAI, with 800 million weekly users, struggled to convert a fraction to Atlas. Combined, AI browsers account for less than 1% of the global market. Installing a new, unfamiliar browser with many new functions is a significant ask.
2. Security Risks (Prompt Injection): AI browser agents, capable of reading, clicking, and filling forms, are vulnerable to "prompt injection." Hidden instructions on webpages (e.g., same color as background) can manipulate the AI into sharing private details, delivering malware, or executing malicious code. OpenAI admitted that prompt injection is unlikely to be fully "solved," and Gartner advised companies to block AI browsers until safer versions emerged.
3. Legal Challenges: Major websites, like Amazon, sued AI browser companies. In November 2025, Amazon sued Perplexity over Comet's shopping agent, alleging it disguised itself as a regular browser while accessing user accounts. While an initial ruling went against Comet, an appeals court overturned it, finding the user, not Perplexity, was accessing Amazon. The legal battle continues.
4. Lack of Unique Value Proposition: For most users, AI functions didn't necessitate a new browser. Existing AI tools like ChatGPT or Claude could already perform searches effectively. The added value of a separate AI browser, rather than integrating AI into an existing browser, was not high enough. Microsoft moved Copilot's browsing features into Edge, and OpenAI integrated Atlas's features into ChatGPT and a Chrome extension, demonstrating this shift.

The Unexpected Winner: Google Chrome

Ironically, the one AI browser that succeeded was Google Chrome. Google integrated Gemini into Chrome, launched "Auto Browse" in January 2026 (allowing the AI to click, scroll, and fill forms with user confirmation), and expanded Gemini to Android users, introducing agents that could operate Chrome using saved passwords. Chrome won because it already had the users.

This phenomenon aligns with the "9x effect" from Harvard Business School, which suggests users overvalue what they already have by 3x, and companies overvalue their new product by 3x. Therefore, a new product needs to be 9x better for users to switch. AI browsers failed to provide this significantly superior experience.

The Monopoly Paradox

The story of AI browsers is intertwined with Google's long-standing legal battle over its search monopoly. Google pays billions annually to maintain its default search engine status on various browsers, including Safari and Firefox. At one point, courts considered forcing Google to sell Chrome. However, in an unexpected turn, the rise of AI browsers became a reason Judge Amit Mehta allowed Google to keep Chrome in September 2025. He believed generative AI had "changed the course of this case" and that AI companies could now compete. Yet, within a year, Atlas was gone, and Chrome had become the dominant AI browser, making Google even harder to beat.

Even Microsoft's broader Copilot strategy faced challenges, with users finding its integration into Windows annoying and its inability to perform basic tasks frustrating.