Your browser just got a lot smarter — and a lot nosier.
Microsoft rolled out one of the most ambitious Edge updates in years on May 13, transforming Copilot from a chatbot sidebar into a memory-equipped, context-aware AI companion. It reads your tabs. It remembers your preferences across sessions. It clicks buttons and fills out forms on websites for you.
The reaction has been swift and polarized. PCWorld called parts of it “replacing browser history with AI slop.” CNET flagged Microsoft’s own warning to “avoid using sensitive or personal information” with the agentic features. Meanwhile, early adopters on Reddit are raving about an AI that pulls up quarterly revenue figures from Power BI reports they viewed weeks ago — without opening a single tab.
Here’s what actually shipped, and why it matters way beyond your browser choice.
The Four Big Features
Long-term memory is the headliner. Enable it, and Copilot remembers your preferences, projects, and past conversations across sessions. Planning a trip to Japan? Weeks later, it recalls your hotel preferences and dietary restrictions without you repeating anything. Microsoft says personal data is processed locally by default, with only derived memory embeddings uploaded to encrypted cloud storage.
Tab-aware answers let Copilot reason across your open tabs — with permission. Microsoft demoed a user with a competitor analysis doc, a Slack thread, and a Google Sheet open, and Copilot synthesized data from all three with a single prompt. The system uses Microsoft’s Phi-4 vision model to understand both text and page layout, even on dynamically loaded sites.
Agentic browsing is the most ambitious and controversial feature. Copilot takes actions on web pages: filling forms, clicking buttons, navigating multi-step flows, all while you watch. Every action requires explicit confirmation, and a new Agent Console sidebar shows a live feed with pause and undo buttons.
Journeys replaces traditional browser history with AI-organized topic cards that summarize your past browsing into resumable projects. And this is where things get ugly.
When AI Summaries Replace Your Actual History
PCWorld’s Mark Hachman didn’t mince words. The problem with Journeys: it replaces your browsing history — the actual links to sites you visited — with AI-generated summaries that don’t link back to anything.
Instead of quickly revisiting a specific page you found two days ago, you get a Copilot-generated summary of what it thinks you were researching.
“Now I have to stop, search, and try to find what I was looking for previously,” Hachman wrote. “How horribly unproductive that is!”
The irony is rich. A feature designed to boost productivity may actually tank it for power users who rely on browser history as a precise recall tool. Microsoft confirmed Journeys is opt-in, but the direction is clear — the company wants AI to mediate your relationship with your own browsing data rather than giving you raw access to it.
The Privacy Elephant in Every Tab
Microsoft emphasizes everything is opt-in and controllable. You can disable memory per-site, per-topic, or entirely. There’s a Memory Dashboard for viewing and deleting stored items. Agentic browsing shows a visible blue outline when active and requires confirmation before acting.
But the details tell a different story.
History-based suggestions are on by default. Tab data gets sent to cloud servers for processing. And Microsoft’s own support page warns users to “be careful to avoid using sensitive or personal information” with agentic browsing — including financial data, Social Security numbers, and medical records.
Read that again: Microsoft built a feature that fills out forms on your behalf, then warns you not to let it near anything important.
There’s also the prompt injection problem. Microsoft acknowledged that “experimental agentic browsing can make mistakes or be affected by hidden instructions on web pages.” Translation: a malicious website could potentially manipulate what Copilot does on your behalf. Microsoft mitigates this with sandboxing and confirmation steps, but for enterprise IT administrators, this is nightmare fuel.
The Browser Is Now the AI
The retirement of Copilot Mode is telling. Previously, AI features lived in a distinct mode you could toggle. Now, Copilot is simply part of the browser. There’s no “AI mode” because the entire browser is the AI mode.
This mirrors the broader industry trend. Google Chrome already offers AI-powered history search. Apple’s Safari is quietly integrating on-device AI. But Microsoft is making the boldest bet: that users want their browser to be an active agent, not a passive window.
The browser — where most knowledge workers spend 60-80% of their day — was always going to be the next AI battleground. Microsoft just fired the biggest shot yet.
What You Should Do Right Now
If you’re among Edge’s 300 million users:
Check your settings immediately. Review what Copilot features are enabled. Journeys and history-based personalization may be on by default depending on your region. The Memory Dashboard is your friend — use it.
Agentic browsing is useful but risky. For low-stakes tasks like comparing products or booking restaurants, it could save real time. For anything involving personal data, financial information, or sensitive work accounts — keep it manual.
Collections is dying. If you relied on it to save and organize tabs, start migrating now. Microsoft indicated Collections will be killed later this year.
Evaluate your browser choice. Firefox, Brave, and Arc are all making different bets on the AI-browser relationship. If Microsoft’s vision makes you uncomfortable, now’s the time to look around.
The Tension Nobody Has Resolved
Microsoft’s Edge update previews where all software is heading. Memory, context awareness, autonomous action — that’s the central thesis of the agentic AI era. Every major platform is converging on this vision.
But the features that make AI most useful are the same features that make it most dangerous from a privacy standpoint. Microsoft’s answer is opt-in toggles and dashboards. Whether that’s sufficient for a tool that lives inside your most intimate digital space remains an open question.
The days of the browser as a dumb window to the web are over. The only question left: will we maintain meaningful control over what it sees, remembers, and does?