Google Resolves Discover Bug: Clarity Restored to AI Summary Links

In a move that highlights the ongoing volatility of Google’s search ecosystem, the search giant has officially addressed a concerning UI inconsistency within its Discover feed. Users and researchers recently identified a bug that effectively "broke" the functionality of source links within AI-generated summaries on Google Discover, preventing users from navigating to the original publishers.

Rajan Patel, Vice President of Engineering for Search at Google, confirmed that the issue was the result of a malfunctioning internal experiment. The resolution serves as a reminder of the delicate balance Google must maintain between integrating generative AI features and ensuring that traffic continues to flow to the content creators that power its ecosystem.

The Core Issue: Disabling the "Source" Pathway

For publishers and SEO professionals, the ability for users to click through from an AI-generated summary to an original source is vital. When Google Discover displays an AI summary, it typically includes "Source" buttons or icons that, when tapped, lead the user to a broader feed or the specific article itself.

The issue, first identified by Google Discover researcher Damien Andell on October 23, involved a specific internal flag identified as disable_aio_sources_button_tap. As the name suggests, this flag effectively rendered the source links inert. Users who attempted to click these links found that the UI provided no response, effectively siloing the user within the AI summary card and preventing them from reaching the publisher’s website.

This discovery sent a ripple of concern through the SEO community. Because Google’s generative AI features already raise questions regarding click-through rates (CTR) and user retention, any UI change that prevents navigation to the source is viewed as a direct threat to publisher traffic.

Chronology of the Discovery and Resolution

The timeline of this incident illustrates how quickly search features can change under the hood of Google’s testing framework.

• October 23, 2024: Researcher Damien Andell publicizes the existence of the disable_aio_sources_button_tap flag on X (formerly Twitter). He provides visual evidence showing that clicking "Source" icons beneath AI-generated video summaries on Discover resulted in no navigation.
• October 23, 2024: Search industry expert Lily Ray amplifies the issue, bringing it to the attention of high-level Google representatives.
• October 28, 2024: Rajan Patel, VP of Engineering for Search at Google, responds directly to Lily Ray on X. He classifies the issue as a "bug in a Discover experiment" and confirms the team has deactivated the test, restoring standard functionality.
• October 29, 2024: Damien Andell confirms that the initial "broken" experiment has been removed. However, he notes that it has been replaced by a new, mysterious experiment labeled "aio tap falls through to hero."

The rapid turnaround—taking less than a week from initial report to official acknowledgment and rollback—demonstrates that Google is monitoring community feedback regarding its AI integrations with significant urgency.

Understanding the "AIO" Context

The term "AIO" refers to AI Overviews and AI-generated summaries, which are becoming ubiquitous across Google Search and Discover. These summaries are designed to synthesize information from multiple web sources. While Google frames these as a value-add for users, they create a "middleman" dynamic between the reader and the publisher.

When a user taps an AI summary, they are looking for information. If the source links are disabled, that user is effectively locked out of the publisher’s site. Given that Google’s Search Console now tracks "Discover Generative AI" traffic, the behavior of these UI elements is of extreme interest to site owners.

According to official Google support documentation, the "Generative AI" report in Search Console tracks impressions. Crucially, these impressions reflect when a user sees a link to a site within an AI summary. However, the metrics do not explicitly differentiate between a user reading the summary and a user clicking through to the site. If the "Source" button is disabled, the value of that "impression" for the publisher drops to near zero.

Implications for Publishers and the SEO Ecosystem

The "broken" button incident is more than just a minor technical glitch; it highlights three critical areas of concern for the modern web:

1. The Fragility of Traffic Referrals

Publishers rely on predictable behavior from search interfaces. If a simple internal test can inadvertently disable navigation to a website, it exposes how much power Google holds over referral traffic. When these "bugs" occur, they underscore the need for publishers to diversify their traffic sources beyond organic search and Discover.

2. Transparency in AI Testing

Google conducts thousands of experiments annually. However, when an experiment directly impacts how a user interacts with a third-party website, the community often demands more transparency. The fact that an internal flag could break navigation without an immediate public disclosure suggests that Google’s internal testing environment may sometimes prioritize data collection over user experience (UX) consistency.

3. The "Hero" Experiment and Future UX

With the introduction of the "aio tap falls through to hero" experiment, it is clear that Google is still iterating on the ideal way to present AI sources. "Hero" usually refers to the primary content piece or the main landing page associated with an AI summary. Google is clearly searching for a balance between keeping the user engaged in the summary and successfully funneling them to the publisher’s content.

A Call for Vigilance

For SEOs and digital marketers, the takeaway from this incident is a need for constant monitoring of the "Generative AI" landscape. As Google continues to refine how AI summaries appear in Discover, standard auditing procedures should include:

• Regular UX Audits: Don’t just check your rankings; check your presence in AI summaries. Are the links working? Are they leading to the correct pages?
• Search Console Monitoring: Watch the "Generative AI" report in Search Console. If you see a sudden, inexplicable drop in impressions or clicks, it may not be a content issue—it could be a UI experiment or a bug similar to the one identified by Andell.
• Community Engagement: Platforms like X have become the primary battleground for identifying these issues. Following researchers who actively monitor Google’s code and UI changes can provide an early warning system for publishers.

Conclusion: A Delicate Balance

Google is currently in a high-stakes transition period. As it pivots from a traditional "link-based" search engine to a "generative-first" information hub, the underlying mechanics of how the web is accessed are changing.

The incident involving the disabled source clicks serves as a cautionary tale. While Rajan Patel and the Google engineering team were quick to rectify the situation, the episode reveals the potential for unintended consequences in a fast-moving AI environment. For the publisher ecosystem, the message is clear: Google’s AI features are still in a state of flux. While Google maintains that these features are designed to help users, the mechanics of these systems must be transparent and functional to ensure the web remains an open, navigable space.

As we look toward the future of search, the industry must remain vocal about the importance of attribution and clickability. Without the ability for users to easily traverse from an AI summary to the source, the relationship between the search engine and the content creator is at risk of fracturing. Google has demonstrated it is listening, but the burden of monitoring these shifting interfaces will continue to fall on the shoulders of the publishers themselves.