Google Says Discover Source Tap Issue Was a Bug

Google says disabled source taps on Discover AI summary cards were caused by a bug in an experiment, and that the test has been turned off. The update was reported by Search Engine Journal on October 27, 2026.

According to the report, Rajan Patel, Google’s VP of Engineering for Search, replied on X that the issue was “a bug in a Discover experiment” and that the normal experience was restored. The report says Damien Andell had first flagged the issue on October 23.

What changed

Before this test, tapping the sources shown under a Discover AI summary opened a feed of articles from those sources. Andell reported that an experiment disabled those taps, which removed one path from an AI summary to the cited sites.

After Patel’s reply, Andell said he checked again on October 27 and found that the earlier experiment had been removed and replaced with another one. The source report says the new experiment was labeled “aio tap falls through to hero,” but it did not explain how that replacement works or where it is running.

What it means for your website

If your site appears in Discover’s generative AI features, this matters because source taps can affect how people move from an AI summary to publisher content. During the reported test, that path appears to have been interrupted for some users, though Google did not say how long the experiment ran or how many people saw it.

For site owners, the practical point is that Discover visibility and Discover traffic paths are not the same thing. The Search Console Discover generative AI report tracks impressions, which the source article says means a user saw your link through Discover’s generative AI. It does not describe source taps as a reported metric.

What to do now

• Check your Discover performance in Search Console, especially the Discover generative AI report, and watch for changes in impressions by URL.
• If you use Discover regularly, test AI summary cards in your own feed and see whether tapping sources opens the article feed.
• Do not assume an impressions change means the same thing as a click path change. Treat visibility and traffic behavior as separate signals.
• Monitor for more detail from Google or from Andell about the replacement experiment, since the current report does not explain how it works.