Watch: AI Mode is Not Live Web
An experiment testing Google's AI Mode suggests it may rely on Google's existing index or cached web data rather than performing live HTTP requests for all URLs.
Transcript
When you use Google's AI Mode to fetch a web page, you might assume it is browsing the live internet in real time. But a recent experiment reveals that this is not exactly how it works.
To test this, a brand-new web page was uploaded to a live server. A standard connection check confirmed the page was live and fully accessible. However, when a Python script was run inside Google's AI Mode to fetch that same new web address, it returned a "not found" error.
The script was then changed to fetch an older, already indexed page on the same server. This time, the AI successfully retrieved the page and its content.
This suggests that Google's AI Mode does not perform a fresh, live request to the target server every time. Instead, it seems to consult Google's existing search index or cache first. If a page is brand-new and has not been crawled yet, the AI acts as if it does not exist and simulates a 404 error, even if the page is actually live.
For developers and researchers, this is a crucial distinction. It means the AI's web access is mediated through Google's existing snapshot of the internet, rather than direct, real-time browsing. If you need up-to-the-second accuracy on a newly published page, the AI might tell you it simply is not there.