Google May Make AI Mode in Search Available by Default

Red Dog Security

09 Sep 2025 — 2 min read

Google is preparing to give users more control over its AI-powered search experience by allowing “AI mode” to be set as the default—potentially replacing traditional link-based results.

What AI Mode Does

AI mode is a version of Google Search that uses large language models (LLMs) to generate summaries of information from across the internet. Instead of directing users to external websites, it delivers answers directly on the search page.

The feature can handle complex questions, process images, summarize articles, and even generate tables, graphs, charts, or code. For now, AI mode is optional, accessible via a tab to the left of “All,” and available in English across 180 countries and territories.

Shifting Toward Default

Speculation about a wider rollout grew last week when Google AI Studio Product Lead Logan Kilpatrick suggested on X that AI mode could soon become the default. Shortly after, Robby Stein, Google’s Vice President of Product for Search, clarified that the company does not intend to force the feature on all users.

Instead, Stein said Google will make it easier for those who prefer AI mode to enable it as the default. In such cases, traditional link results will be hidden, though they can still be accessed under the “Web” tab, located at the far end of the results panel.

Business and Industry Concerns

While Google maintains there are no immediate plans to activate AI mode by default for everyone, observers note that such a shift could be imminent. Internally, engineers are weighing how the change might impact the company’s advertising model.

Google has already begun testing ads in AI mode and in AI-generated overviews, offering these formats to partners. But uncertainty remains within the digital marketing industry about how advertising will function if classic search results—and the billions of clicks they generate—are displaced.

Currently, Google controls around 90% of the search market and serves as a primary traffic source for publishers worldwide. Yet publishers receive no compensation when their content is summarized by Google’s AI. Instead, Google argues that AI summaries deliver more “high-quality” clicks, though it has not released supporting data.

Pushback From Publishers

Independent research suggests users are less likely to click on links when presented with AI summaries. Media reports indicate that some publishers are now considering forming alliances to defend against what they describe as an existential threat: the integration of AI into search.

The s1ngularity Attack Affected 2,180 GitHub Accounts

Specialists from Wiz have published new details on the recent s1ngularity attack targeting the open-source build system NX. The incident exposed data from 2,180 GitHub accounts and compromised more than 7,200 repositories, making it one of the largest supply chain breaches seen this year. What Is NX? NX

iCloud Calendar Is Being Exploited to Send Phishing Emails from Apple’s Servers

Invitations sent through iCloud Calendar are being weaponized to deliver phishing emails disguised as purchase notifications—sent directly from Apple’s own mail servers. This tactic significantly increases the chances of bypassing spam filters. A Familiar Scam with a New Twist According to Bleeping Computer, a reader recently shared a

Hacked npm Packages Downloaded More Than 2.6 Billion Times Weekly

Researchers are warning of what may be the largest supply chain attack in history. Attackers injected malware into several of the most widely used npm packages—collectively downloaded more than 2.6 billion times each week. The breach was made possible when hackers compromised a maintainer’s account through a

Unpatched Vulnerability Found in TP-Link Routers

TP-Link has confirmed a previously unknown zero-day vulnerability affecting several of its router models. The company said it is investigating the issue and preparing security updates. Discovery and Technical Details The flaw was discovered by an independent researcher using the alias ByteRay, who first reported it to TP-Link on May