
Image credit: Abondance
Google is testing a new artificial intelligence-generated summary feature in its Discover feed, potentially impacting web publishers by providing topic overviews before linking to their content.
The new feature, called “Dive deeper,” offers users a concise AI-generated summary of a topic, which initially does not directly link to a publisher’s specific article, according to reports.
Technology reporter Barry Schwartz first reported the test on Sept. 18, 2026, after Google‘s Robby Stein announced the initiative on X.
Google stated its objective for the new feature is to simplify topic exploration and enhance the discovery of related web content within Discover.
The initial phase of the “Dive deeper” test will concentrate on video content, with Google planning to experiment with various design iterations in the coming weeks, Stein said.
Publishers have expressed significant concerns that providing comprehensive context through AI summaries could diminish user incentive to visit their websites, potentially leading to a reduction in traffic.
A more extensive version of the “Dive deeper” overview includes links to related stories, community reactions, and original reporting, but this appears after the initial AI summary.
The introduction of this intermediary AI layer has heightened worries among content creators regarding their ability to monetize their work and maintain direct audience engagement.
Google Discover, a personalized content feed, has become a notable source of traffic for many news and content publishers, making any changes to its functionality a point of close observation within the industry.
The company has not yet provided a timeline for a broader rollout of the “Dive deeper” feature beyond the current testing phase.
Source: Abondance
Written by
Palumbo Angela
Angela Palumbo, Senior Editor at Rabbit Rank since 2023, holds a bachelor's in communications. She focuses on fact-checking and simplifying complex topics while also leading strategy for the news department.
Keep reading
Related Articles

Google releases Gemini 4 Argon AI to Pro, Ultra subscribers
Google’s Gemini 4 Argon AI model is now available for Google AI Pro and Ultra subscribers, developers, and ent...

SEO Experiments Often Fail Due to Lack of Scientific Rigor
SEO experiments frequently fail to produce useful insights because they neglect scientific principles, leading...

AI Brand Presence Low Correlation with Citations, Study Finds
Pierre Sauvé’s study reveals low correlation between brand presence in AI models and citation rates, highlight...