Microsoft is folding Copilot deeper into Edge, making the browser less like a place to visit AI and more like the place where AI follows the work.
Microsoft's latest Edge update is not really about a smarter sidebar. It is about distribution. On May 13, the company moved a fuller version of Copilot across Edge on desktop and mobile, giving it permissioned access to tabs, browsing history, past chats, voice, vision and the new tab page. That is a lot of surface area for one assistant.
The shift matters because the AI browser race has been building as if the winner will be whoever launches the most interesting new browser. Microsoft is taking a different route. It already has a browser sitting on Windows machines, available on phones, connected to Microsoft accounts and increasingly tied into Microsoft 365. That means Edge can become an AI browser without asking most users to make a dramatic switch.
According to Microsoft's Edge blog, Copilot in Edge can now reason across multiple open tabs on desktop and mobile with user permission, compare information, surface key details and answer questions without forcing people to hop between pages. It can also use browsing history and prior Copilot chats to give more relevant answers, while long-term memory is now part of the desktop and mobile experience.
The most important change is not any single feature. It is the way Microsoft is moving Copilot from a destination into the workflow itself. The redesigned new tab page brings chat, search and browsing into one starting point. Journeys groups browsing activity into projects with summaries and suggested next steps. Copilot quizzes can turn a page into study material. The writing assistant can generate drafts or revise text where a user is already typing. A podcast generator can turn tabs into something you listen to later.
That is a broader ambition than answering questions about a webpage. Microsoft wants Edge to remember what you were doing, understand what is open, help you produce something and keep the work moving. For a user planning travel, researching vendors or comparing software, the browser starts to act less like a passive container and more like an operating layer for everyday decisions.
The limits are worth paying attention to because they show how early and uneven this rollout still is. Journeys on desktop is available only in English markets, while Journeys on mobile is U.S.-only. Writing Assistant is U.S.-only. Browse with Copilot, the agentic feature formerly known as Copilot Actions in Edge, is limited to Microsoft 365 Premium subscribers in the U.S. on desktop. Podcasts are available in English markets and require a Microsoft account to generate.
Microsoft is also retiring Copilot Mode as standalone branding. That is a small naming change with a larger message behind it. Copilot is no longer being treated as a special mode inside Edge. It is being absorbed into Edge itself.
Consent is now the product risk
The privacy question is no longer theoretical. If an assistant can use open tabs, history, past chats and screen context, the value comes from seeing more of the user's work. The risk comes from the same place. Microsoft says these features require permission, that users can choose which experiences to enable and that Copilot follows the company's privacy standards. It also says users can disable context clues and personalization settings.
Browse with Copilot makes the tradeoff especially clear. Microsoft's support materials say the feature can select, type, scroll and navigate inside a tab while the user watches. It can access cookies for sites where the user is already signed in, and it can see open tabs in the current browser window. Microsoft says it cannot access saved passwords, wallet information or autofill data, and that users can interrupt or take control at any time.
Those safeguards matter, but they do not remove the operational risk. Microsoft warns that agentic browsing can be affected by prompt injection, hidden malicious instructions on webpages, unintended actions and financial exposure. That is a candid admission of where AI browsing is heading. The more useful the assistant becomes, the more it needs boundaries that ordinary users can actually understand.
This is where startups face a hard problem. Perplexity has Comet. The Browser Company has Dia. Others are trying to build browsers around AI from the ground up. Their pitch is clarity: start fresh, design the whole experience around the assistant and make browsing feel conversational from the first click. That can be powerful for early adopters.
Microsoft's advantage is that it does not need early adopters to carry the market. It can put AI browsing in front of people through Edge, Windows, mobile apps and Microsoft 365 subscriptions. Even if some users ignore it, the default position is valuable. In consumer software, repeated exposure can be just as important as product purity.
For businesses, the takeaway is more practical than philosophical. The browser is becoming a place where AI reads, remembers and acts. That will change how people research products, compare vendors, write documents, study material and move across websites. It will also force companies to think about how their sites behave when the visitor is partly human and partly agent.
The next phase of the AI browser race will not be judged only by clever demos. It will be judged by trust, permissions, speed and whether users feel in control when the assistant starts touching the web on their behalf. Microsoft has put Edge in the middle of that contest. Now it has to prove that default access can become daily habit without turning convenience into discomfort.
Also read: Meta is bringing private AI chats to WhatsApp. • AIDC-AI brings cheaper visual reasoning to open multimodal AI • Notion wants to become the control room for AI agents
