Google Confirms "Disabled" Discover Source Clicks Were An Unintended Bug

In a move that highlights the ongoing volatility of AI-driven search experiences, Google has officially addressed concerns regarding the functionality of source links within its Discover feed. Users and researchers recently observed that AI-generated summaries on Google Discover had lost their ability to navigate to source material, sparking concerns about the future of traffic attribution for publishers. Rajan Patel, Google’s VP of Engineering for Search, confirmed that the issue was not a deliberate policy shift but rather an unintended consequence of an internal experiment.

The clarification provides a momentary sigh of relief for SEO professionals and publishers who closely monitor how Google’s evolving AI features—often referred to as AI Overviews or AI-powered summaries—impact their visibility and referral traffic.

The Chronology of the Disruption

The issue first came to light on October 23, 2024, when Google Discover researcher Damien Andell identified a troubling trend: users were no longer able to click on the "Sources" links provided beneath AI-generated summaries in their Discover feeds.

October 23: Discovery of the "Disable" Flag

Andell, who frequently analyzes backend changes in Google’s ecosystem, discovered that the functionality had been stifled by an internal flag labeled disable_aio_sources_button_tap. Prior to the implementation of this flag, tapping the "Sources" section would typically open a curated feed of articles from the cited publications, providing a seamless pathway from a summarized AI insight to the original journalistic content.

October 23–27: Community Concern and Public Discourse

The discovery quickly caught the attention of prominent industry figures, including SEO analyst Lily Ray. The concern was immediate: if users are presented with a summary of content but are prevented from clicking through to the source, the publisher essentially loses the "last mile" of traffic, rendering the AI summary a terminal point for the user’s journey rather than a bridge.

October 28: Official Clarification

Following public inquiry, Rajan Patel, VP of Engineering for Search, addressed the matter directly on X (formerly Twitter). Patel stated, "Hey Lily — this was a bug in a Discover experiment. We’ve turned off the experiment, and the experience works as intended again."

The Aftermath

By October 29, Andell confirmed that the initial experiment had been removed. However, he noted that it had been replaced by a new, mysterious experiment labeled "aio tap falls through to hero." This suggests that while the "bug" was resolved, Google remains in an active phase of testing how users interact with AI summaries in the Discover environment.

Understanding the Google Discover Ecosystem

Google Discover is a highly personalized feed that presents users with content based on their search history and stated interests. Unlike traditional search, which is intent-driven, Discover is discovery-driven, making it a critical traffic driver for news publishers and content creators.

The Role of AI Summaries

In recent months, Google has integrated Generative AI into Discover to provide concise summaries of videos and long-form articles. The goal, according to Google, is to help users quickly grasp the essence of a topic. However, these summaries operate on a precarious balance: they must provide enough information to be useful, but they must also drive traffic back to the original source to ensure the sustainability of the web ecosystem.

The Importance of "Source Taps"

The "Sources" link acts as a primary CTA (Call to Action) for the publisher. When a user taps these sources, they are typically taken to a deeper collection of content from that publisher, increasing session duration and engagement. When these taps are disabled, the publisher is stripped of the opportunity to convert a casual "feeder" reader into a loyal site visitor.

Official Responses and Transparency

The interaction between industry analysts and Google executives highlights the "black box" nature of modern search development. Patel’s rapid response was lauded by the SEO community for its clarity, yet it also underscored the lack of documentation surrounding these experiments.

The "Bug" vs. "Feature" Dichotomy

In the world of software development, the line between an experiment and a bug is often thin. A feature intended to measure user behavior can often have side effects that make the experience feel "broken" to the end user. Patel’s confirmation that this was a "bug" suggests that the team did not intend to completely block the functionality; rather, the code responsible for the experiment likely overreached or failed to trigger the proper redirection logic.

Lack of Public Disclosure

Despite the resolution, many publishers remain frustrated by the lack of transparency regarding Google’s testing cycles. Because these experiments happen in real-time, often affecting only a fraction of the global user base, publishers rarely have a way to verify if their traffic drops are due to algorithmic changes, site errors, or Google’s internal testing.

Implications for Publishers and SEO

This incident serves as a microcosm for the broader anxieties currently facing the publishing industry. As Google continues to lean into "Answer Engines," the value of a referral is being redefined.

Traffic Attribution and Search Console

Google’s Generative AI Report in Search Console is the primary tool publishers use to gauge their performance in AI-driven features. Currently, the report tracks "Impressions"—which counts how many times a user sees a link in a summary—but it does not provide granular detail on "Source Taps."

This creates an information asymmetry. If a publisher sees an increase in AI impressions but a decrease in actual clicks, they may struggle to diagnose the cause. Is the content failing to entice a click, or is the interface itself inhibiting the user? The recent bug incident highlights that sometimes the barrier is entirely technical, rather than content-related.

The Shift Toward "Zero-Click" Environments

The "aio tap falls through to hero" experiment, which replaced the buggy version, signals that Google is still refining the user experience. Whether this means future summaries will redirect users directly to the publisher’s "hero" content rather than a list of sources remains to be seen. For publishers, this uncertainty necessitates a diversification strategy:

1. Monitor Referral Traffic: Don’t rely on aggregate traffic numbers. Use advanced segmentation to isolate traffic coming specifically from Google Discover vs. Search.
2. Analyze CTR Trends: If you notice a sudden, inexplicable drop in CTR (Click-Through Rate) from Discover, investigate whether it aligns with recent news regarding AI summary updates.
3. Engage with the Community: As seen with the reports from Damien Andell and Lily Ray, industry researchers are often the first line of defense in identifying these technical anomalies.

Looking Ahead: The Future of AI in Discover

As we move further into 2025 and 2026, the integration of Generative AI into mobile feeds will only become more sophisticated. The "bug" in this specific experiment is a reminder that even the most powerful tech companies in the world are constantly iterating, often at the expense of stability.

For the SEO community, the lesson is clear: keep a close watch on the interface elements of your top-performing channels. When Google experiments with its UI, the impact on your business can be immediate and significant. Whether the "aio tap falls through to hero" test will eventually become a permanent feature or be relegated to the history books remains an open question.

For now, the ability to click through to a source remains intact, ensuring that the vital link between content creators and their audience stays open—at least until the next experiment begins.

How to Monitor Future Changes

Publishers who are concerned about future experiments should:

• Audit Regularly: Periodically check your own Discover feed to see how your content is being presented.
• Follow Industry Watchers: Accounts like those of Damien Andell and other SEO experts provide early warnings on backend changes.
• Review Search Console: Watch for significant shifts in "Generative AI" impression data, even if the click-through data remains sparse.

In the fast-paced world of search engine optimization, vigilance is the only constant. While Google’s quick fix to the source-tap bug was a positive sign of accountability, it serves as a stark reminder that the ground beneath digital publishers is constantly shifting as AI continues to reshape the user journey.