Google says the disabled source taps on Discover AI cards were a bug

Google says the disabled source taps on Discover's AI summary cards were "a bug in a Discover experiment". Google has turned the experiment off. The words are Rajan Patel's. He is Google's VP of Engineering for Search. Search Engine Journal's Matt G. Southern reported it from a reply Patel posted on X Monday evening.

The test was spotted on 23 October by Damien Andell, who researches Google Discover. The AI summary cards in Discover carry a row of source icons at the bottom. Tapping one usually opens a list of the articles the summary drew from. Andell found that a feature flag called disable_aio_sources_button_tap stopped the row from opening anything.

What Google said

Patel was replying to Lily Ray, who had quote-posted Andell's report the morning of 23 October. Patel wrote: "Hey Lily, this was a bug in a Discover experiment. We've turned off the experiment, and the experience works as intended again."

The reply does not say what the experiment was testing. It does not say how long it ran. It does not say how many people saw the disabled taps. Google has not published anything else about it.

A different experiment replaced it

On 27 October Andell posted that he had checked again. The first experiment was gone, and a new one named "aio tap falls through to hero" had taken its place. Andell's post did not explain what the new experiment does, or where it is running.

So the source taps work again, and Discover is still being tested in that spot on live users. The flag name hints that a tap now falls through to the hero item. Nobody outside Google knows what that means in practice yet.

What a publisher can check

Open the Google app on a phone. Find an AI summary card in the Discover feed. Tap the source icons at the bottom. If a list of articles opens, the fix has reached that account.

The Discover Generative AI report in Search Console counts impressions on links shown in a Discover AI card. It does not count taps, so a drop in reader taps would not show up in that report. A publisher who wants to track this has to look at analytics for Discover traffic that followed a sources tap.

Keep that measurement running, because the replacement experiment is live. One flag closed the link in a day. The next one could change how a tap behaves, with no word from Google.

Source

• Matt G. Southern at Search Engine Journal, Google: Disabled Discover Clicks Were A Bug In An Experiment.