Microsoft Copilot & Bing: AI Search Updates and Citation Tracker

Going to copilot.microsoft.com now produces a sign-in screen rather than an answer box. We checked on 1st Sep and found no way past it.

The page offers sign-in with a Microsoft, Apple or Google account and nothing else.

Microsoft confirmed it is deliberate. Krishna Madhavan, a principal product manager at Microsoft AI, Bing, told Search Engine Roundtable: "Going forward we are requiring all Copilot users to login."

The same report puts the boundary in Microsoft's own words. Michael Schechter, corporate vice president for search growth and experiences, said: "Copilot Search doesn't require sign in - I'm focused on Bing, so can't speak to Copilot's features."

So Microsoft's AI answers now come in two forms with different front doors.

A Copilot Search answer on a Bing results page stays open to anyone. The Copilot app and website need an account.

The change reaches Bing's home page too, where the Copilot button leads to the same wall. The story was first spotted by Windows Report before Microsoft confirmed it.

For a site owner this is about reach rather than ranking. Every citation Copilot serves now goes to a signed-in user.

The audience is whoever holds a Microsoft, Apple or Google account and chose to sign in.

Nothing about how pages are selected or cited has changed. Microsoft has said nothing about crawling, indexing or the AI visibility reporting in Bing Webmaster Tools.

It also makes the surface harder to measure from outside. Some third-party tools sample AI answers by loading them anonymously.

Those tools can still reach Copilot Search in Bing. They can no longer reach Copilot itself.

Microsoft has published no blog post or help-page note about the sign-in rule. So the only statements on the record are the two quoted above.

This is a sighting, not a Microsoft announcement.

Sachin Patel documented it at Search Engine Watch on 26th Aug with before-and-after screenshots from mobile.

The AI answer at the top of Bing results appeared under an "AI Overview" label. It normally says Copilot.

His own framing keeps the doubt in view: "Whether this is a rebrand test or a simple mislabeling bug isn't clear yet."

Search Engine Roundtable wrote it up the next morning and read it as a test. It noted that "Microsoft is still placing the Copilot logo with the box".

Microsoft has published nothing, and no help page has changed. Its own June wording still calls the surface Copilot Search.

So treat the current name as unchanged until Microsoft says otherwise.

The reason a label test earns a place here is what the label is. Google's product name is "AI Overviews".

If Bing adopted it, the industry shorthand for AI answer boxes would stop separating the two engines. That matters to anyone reporting on "AI Overviews" traffic or citations by name.

Nothing in the sighting suggests the answers themselves changed, only what they are called.

Microsoft expanded the AI Performance report in Bing Webmaster Tools with four new views, all "available in preview globally".

Intents classifies the grounding queries behind your citations into intent categories.

Topics groups them into thematic clusters. Compare overlays two time periods of citation activity.

Citation Share is the headline: your site's share of all citations for a query. Microsoft calls it "designed as an observational metric", not a ranking system.

It exposes no competitor domains, no traffic share and no quality scores.

The June post also restates what the report covers: "Microsoft Copilot, Bing, and select partner AI experiences".

For context, Google's Generative AI reports in Search Console, announced thirteen days earlier, carry impressions only. Bing is reporting the citations themselves, with the retrieval queries behind them.

Ten months after retiring the Bing Search APIs, Microsoft announced their strategic successor: "a suite of AI-native grounding APIs built for the agentic era".

The framing is the story.

"Web IQ is a search engine for AI systems. Where Bing was built to help people search the web, Web IQ is built to help AI agents find the right information, turn it into useful evidence, and use it inside reasoning."

Instead of ranked blue links, Web IQ returns passage-level "structured evidence objects". Microsoft claims sub-165ms p95 latency, measured in its own internal comparisons.

So treat the number as a vendor claim.

For site owners, the post says Web IQ honors robots exclusion protocols and publisher controls. It also says Microsoft is engaging the IETF on AI-era standards for the open web.

Availability is the main caveat. The post ships no product, no pricing and no dates, only an interest form.

Anything you may read about specific preview regions is not in Microsoft's post.

Microsoft ended its named AI browser experiment: "As part of today's update, we're retiring Copilot Mode. With helpful features built directly into Edge, it's now simpler to shape how you browse and get more done."

The features survive the label. Multi-tab reasoning, history-informed answers, memory, Vision and Voice move into default Edge on desktop and mobile.

So do Journeys, Study and Learn, the writing assistant and podcast generation. Some features are gated by market.

The lifecycle: launched 28th Jul 2025 as an opt-in experiment on Windows and Mac. Upgraded in the October Fall Release, retired 13th May 2026.

AI browsers kept churning across the industry; OpenAI's Atlas lasted 292 days.

The same day, Bing published a separate essay on keeping trusted sources visible inside AI answers. The essay is "Keeping Trusted Content Visible in an AI-Powered Search World".

It lists risk detection at query level, public-service announcements and SafeSearch modes.