By SEJ Staff | Updated October 2026

Google has officially addressed widespread industry anxiety regarding a recent functional glitch within its Discover feed. According to high-ranking company executives, the unexpected inability for users to tap or click on source attribution links within Google Discover’s artificial intelligence (AI) summary cards was not a permanent shift in traffic strategy, but rather an unintended software bug tied to a routine product experiment.

The clarification brings a brief sigh of relief to digital publishers, SEO professionals, and content creators who closely monitor referral traffic sources. As AI-generated summaries and overviews continue to claim a larger share of search and discovery real estate, any friction introduced between a user and a citing publisher’s website instantly raises alarms across the digital marketing ecosystem.

1. Main Facts

The core of the issue centers on Google Discover, a heavily personalized content feed that surfaces articles, videos, and media based on a user’s inferred interests rather than active search queries. In recent months, Google has increasingly infused Discover with generative AI features—including AI-driven summaries and contextual cards designed to give users a quick overview of trending topics.

• The Core Issue: For a subset of users, tapping on the designated "sources" or citation links at the bottom of these AI summary cards failed to trigger the expected navigational action. Instead of opening a feed of related articles or routing the user directly to the cited publisher’s website, the buttons were entirely non-responsive.
• The Discovery: The bug first came to light via independent researchers and SEO authorities who actively monitor Google’s interface updates through code flags and front-end testing.
• Google’s Explanation: Rajan Patel, Vice President of Engineering for Search at Google, formally clarified on social media that the disabled clicks were a technical malfunction ("a bug") stemming from an internal Discover experiment.
• Current Status: Google has since deactivated the problematic experiment, and the user interface has reportedly returned to its intended operational state, allowing standard source attribution clicks once again.

2. Chronology of Events

The unfolding of the Discover click-bug narrative occurred rapidly across social media platforms, transforming from an obscure code observation into a major industry talking point over the span of just a few days.

• October 23 (The Initial Report): Google Discover researcher Damien Andell first flagged the anomaly on X (formerly Twitter). Digging into mobile flags, Andell identified a variable labeled disable_aio_sources_button_tap. He demonstrated via screenshots and commentary that this flag effectively cut off the bridge between AI summaries and the underlying source material—preventing users from expanding or viewing the feeds of cited publishers.
• October 23 (Industry Amplification): Within hours of Andell’s post, prominent SEO figures—including Lily Ray—quote-posted the findings, drawing broader attention to the potential implications for publisher traffic. The SEO community immediately began questioning whether Google was quietly testing a closed-ecosystem model for Discover AI summaries that would keep users trapped within Google’s interface.
• October 26–27 (Google’s Official Response): Responding directly to Lily Ray’s thread on Monday evening, Google’s Rajan Patel interjected to clear the air, stating definitively that the behavior was a bug tied to an active experiment and confirming the test had been pulled.
• October 27 (Post-Fix Observations): Following Patel’s confirmation, Andell revisited the Discover interface to verify the changes. He noted that the initial buggy experiment had indeed been retired, though it had quickly been succeeded by a new, distinct test labeled internally as aio tap falls through to hero.

3. Supporting Data and Technical Observations

While Google’s official statement was relatively concise, the technical breadcrumbs left behind by independent researchers offer a fascinating look into how Google tests user experience (UX) variations within its mobile feeds.

The Anatomy of the Bug

Before the test went live, interacting with a Discover AI summary card’s citation icons allowed users to seamlessly traverse from a summarized snippet down into a curated feed of source articles. This mechanic acted as a critical referral pathway, ensuring that users interested in diving deeper could easily transition from Google’s generated text to primary publisher domains.

When the disable_aio_sources_button_tap flag was active, however, this pathway was severed. In visual demonstrations provided by researchers:

1. A media item (such as a YouTube video surfaced via Fox 9) appeared in the Discover feed.
2. An automated AI summary text block sat directly beneath the media asset.
3. A row of favicon-driven links labeled explicitly as "Sources" was positioned at the base of the card.
4. While the icons were visually present, clicking them resulted in a dead end, failing to launch the expected article aggregation feed or outbound link.

The Successor Experiment

Following the deactivation of the source-blocking test, researchers identified a secondary experiment winding its way through the platform: aio tap falls through to hero. While Google has not publicly detailed the exact mechanics of this new test, it suggests that the company is actively iterating on how touch inputs interact with AI-generated summary elements—testing alternative ways for users to navigate from a summary card back to the primary content piece (the "hero" asset) without necessarily relying on traditional multi-source citation carousels.

4. Official Responses

The transparency exhibited in this instance—coming directly from a senior engineering executive—helped quell what could have easily ballooned into a severe panic among publishers.

Rajan Patel’s statement on X was precise:

"Hey Lily — this was a bug in a Discover experiment. We’ve turned off the experiment, and the experience works as intended again."

Despite providing clarity on the specific bug, Patel’s response left several lingering questions unanswered. Google did not disclose:

• The exact geographic regions or device ecosystems (Android versus iOS) impacted by the bug.
• The precise duration the experiment ran before being flagged and shut down.
• The total volume of users or impressions that encountered the non-responsive citation buttons.
• The overarching hypothesis Google was attempting to test by disabling source taps in the first place.

Nevertheless, the rapid acknowledgment and reversal were largely welcomed by a publishing industry that remains hyper-sensitive to any alterations in organic referral traffic architectures.

5. Implications for Publishers and SEOs

The incident highlights a broader, ongoing anxiety within the digital publishing and search engine optimization communities: the delicate balancing act between Google’s deployment of generative AI features and the preservation of publisher referral traffic.

The Role of Search Console and Generative AI Reporting

Google currently provides visibility into AI-driven traffic via the Discover generative AI report in Google Search Console. This dashboard logs instances where user feeds feature links to a publisher’s pages through Discover’s generative AI capabilities, measuring performance primarily through impressions.

However, metrics concerning user interaction—such as actual source taps, dwell time on publisher sites originating from AI summaries, and downstream conversion metrics—remain heavily obscured or segmented differently from traditional organic clicks. When an experiment breaks the clickable pathways entirely, it directly threatens the primary value proposition of appearing as a cited source in an AI summary: user acquisition.

Key Takeaways for Digital Strategists

1. Continuous Monitoring is Essential: Front-end bugs and unannounced UI experiments can severely impact referral metrics overnight. Publishers relying heavily on Google Discover traffic should maintain rigorous daily tracking of referral spikes, drops, and anomalies.
2. AI Integration is Fluid: Google’s UI tests prove that the layout of AI summaries, search overviews, and Discover cards is far from finalized. Expect continuous, iterative testing as Google figures out how users want to consume and interact with AI-synthesized information.
3. Audit Your Own Feeds: As recommended by search analysts, publishers should periodically check their own mobile Discover feeds manually. Tapping on AI summary cards and verifying whether source citations correctly direct traffic to external websites remains a practical, hands-on way to catch interface anomalies before they show up on monthly traffic balance sheets.

Looking Ahead

As Google continues to refine its generative AI integrations across Search, Discover, and beyond, communication channels between platform engineers and the SEO community will remain vital. While this specific instance was confirmed to be a temporary experimental bug, it serves as an important reminder of how rapidly user journeys can be altered—intentionally or accidentally—in the modern search landscape.