AI FactScan
The change looks like an olive branch to website owners whose traffic has been pulled into AI-generated answers. Google appears more willing to send users back to the open web. But has the update fixed the underlying problem? Two questions remain: whether the new links generate meaningful clicks, and whether the sources selected by AI actually support its answer.
What changed in Google AI Mode?
Since AI summaries arrived in search, publishers and website owners have faced a growing problem: zero-click searches. Users can read an AI-generated summary on the results page and leave without visiting the original website. That loss of traffic threatens the businesses that fund original reporting and useful web content.
Google says AI Search will now show inline links more frequently and more prominently, with improved cards for related reading. Publicly, the goal is to help users explore the web. The change also responds to mounting criticism over traffic, market power, and the use of publishers' work.
The obvious question is whether people will actually click those links.
Does Google disclose click data for AI Mode and AI Overviews?
Google has promoted the new links as easier to click, but it still does not provide website owners with specific click data for AI Mode or AI Overviews in Search Console.
That leaves publishers and marketers unable to answer two basic questions:
- How often did an AI answer cite or display their article?
- Did users stop at the AI summary, or click through to the website?
As long as click performance remains hidden, Google's claims about sending traffic to the web are difficult to verify. For publishers that rely on measurable results, more visible links are not enough on their own.
Can you trust the sources selected by AI Search?
For search users, the larger issue is not how many links appear. It is whether those links actually support the AI summary.
It is easy to assume that a link shown by Google AI is a carefully selected, authoritative source. Recent research and real-world cases suggest otherwise.
Even major media outlets can be fooled by AI-generated content
A social media account recently posted an AI-generated fake selfie of actor Mia Goth. Entertainment outlets treated it as real and published stories about it. Once a false claim enters a recognizable media site, Google AI may collect the resulting coverage and summarize it as fact.
AI Search and traditional search rely on very different sources
A study comparing Google Search, AI Overviews, and Gemini found an average source-overlap score below 0.2. AI summaries were more likely to draw from forums, general-interest media, or Google's own ecosystem instead of original official documents or authoritative research.
A credible link may still fail to support the AI's claim
Research has also found that more than 10% of statements in AI Overviews were unsupported or inconsistent with the cited pages. Even when the domain itself is credible, an AI summary can omit context, combine unrelated passages, or attach a source to a claim the page never made.
For high-stakes searches involving schoolwork, research, housing, money, health, or legal decisions, AI Mode should never be treated as the only authority.
How to check AI citations with AI FactScan
You do not need to open every link one by one. AI FactScan can help you run a quick first check and identify the sources that deserve closer attention:
- Check the grades: See whether the answer relies on official institutions, original research, established references, forums, or content farms.
- Spot repeated domains: An answer may look well sourced while citing several pages from the same website.
- Verify the support: Open the key sources flagged by AI FactScan and confirm that the original page actually supports the AI's claim.
More links do not make an answer trustworthy
Adding more links to Google AI Mode is a step in the right direction. But Google still has work to do on traffic transparency and the accuracy of AI-generated summaries.
An AI summary can help you build a quick map of a topic. It should not be treated as the final word. The next time Google AI presents a polished answer with a long source list, use AI FactScan to inspect the source quality before you rely on it.