Google’s introduction of AI Overview has fundamentally altered the search experience for millions of users worldwide, generating significant controversy and user frustration since its wider rollout in 2024. While the feature was designed to provide quick, synthesized answers directly on search results pages, the reality has proven far more complicated, with users encountering inaccurate information, reduced website traffic for publishers, and an experience that many find cluttered and counterproductive. Importantly, Google has made clear that there is no simple, official “off” switch for AI Overviews—the feature cannot be disabled in traditional settings menus, leaving users to employ various workarounds and alternative approaches to reclaim their preferred search experience. This comprehensive analysis explores the multiple methods available to users seeking to disable or avoid AI Overviews, examining both desktop and mobile solutions, browser-based tools, alternative search engines, and the broader ecosystem of implications surrounding Google’s aggressive integration of artificial intelligence into its core search product.
Understanding Google AI Overviews and Their Purpose
Google AI Overviews, formerly known as Search Generative Experience (SGE) during its experimental phase, represent a significant departure from the search engine’s traditional approach to presenting information. Introduced as a Search Generative Experience at Google I/O in May 2023, the feature was rebranded as AI Overviews and officially launched to all United States users in May 2024. The feature functions by generating artificial intelligence-synthesized summaries of search results, which appear prominently at the top of the search results page, above the traditional “blue links” that users have relied upon for decades. These AI-generated summaries attempt to condense relevant information from multiple sources into concise, coherent answers designed to satisfy user queries without requiring them to navigate through multiple websites. Google positioned this innovation as a natural evolution of search, arguing that it would save users time and provide more comprehensive information by synthesizing knowledge from diverse sources across the web.
The underlying technology powering AI Overviews leverages a custom version of Google’s Gemini artificial intelligence model, specifically tailored for the search environment. When a user enters a query, the system scans Google’s index and Knowledge Graph to identify relevant facts and information, then uses machine learning algorithms to synthesize this content into a coherent summary. The AI can break complex queries into multiple sub-questions through a process known as query fan-out, allowing it to provide more nuanced and comprehensive responses than would be possible through simple keyword matching. Google’s implementation includes prominent links to source content within the summaries, allowing users to access the original articles if they wish to learn more. By August 2024, Google had expanded AI Overviews to several additional countries including the United Kingdom, India, Japan, Brazil, Mexico, and Indonesia. As of May 2025, the feature became available in over 200 countries and territories, supporting more than 40 languages, making it one of Google’s most widely distributed experimental features ever.
Why Users Want to Disable AI Overviews
The motivation for disabling AI Overviews extends far beyond mere user preference; it encompasses serious concerns about accuracy, reliability, privacy, and the fundamental structure of the internet itself. When Google AI Overviews first rolled out at scale, the feature became infamous for generating confidently incorrect information that ranged from laughable to genuinely dangerous. One early and widely-publicized example involved the system recommending that users add glue to pizza sauce to prevent cheese from sliding off—advice apparently sourced from a Reddit joke that the algorithm treated as legitimate information. Beyond this notorious example, users have documented instances where AI Overviews confidently assert false information, create entirely fabricated idioms and phrases, provide harmful health advice that contradicts established medical consensus, and present offensive or fringe viewpoints as if they were mainstream understanding. The system has struggled particularly with what researchers have termed “false premise” searches—when users search for things that don’t exist or shouldn’t exist, the AI sometimes attempts to provide helpful-sounding information about non-existent things rather than acknowledging that the premise is flawed.
Beyond accuracy concerns, users have expressed frustration with AI Overviews as a matter of search experience and quality. The feature occupies substantial screen real estate, particularly on mobile devices where space is limited, pushing traditional search results further down the page. For users seeking comprehensive information or wanting to explore multiple perspectives on a topic, the AI-generated summary can feel reductive and insufficient, obscuring the diverse range of sources that Google’s traditional search results would have displayed. A significant portion of users report that they prefer to review multiple sources and make their own judgments rather than relying on an AI-synthesized version of the truth. Additionally, many researchers, academics, and professionals whose work relies on finding and evaluating original sources find AI Overviews to be actively counterproductive, as the feature discourages clicking through to read primary sources and understand the nuance and context behind information.
Privacy-conscious users have raised additional concerns about AI Overviews, noting that the feature is trained on content from across the internet, often without explicit permission from content creators, and the system’s training process raises questions about fair use and the proper compensation of original content creators. Users also report frustration with what they perceive as Google’s aggressive and deceptive marketing of the feature, constantly pushing them toward AI experiences even when they have expressed a preference for traditional search results. The feature’s persistent reappearance after users believe they have disabled it, combined with its integration across multiple Google products and services, has created a sense among users that they are being forced into an AI-first paradigm regardless of their actual preferences.
The Official Google Position on Disabling AI Overviews
Google has been remarkably clear and consistent in its official messaging regarding the ability to disable AI Overviews: there is no official method to turn them off entirely. When users navigate to Google Search’s settings and Search Labs, they will not find an “off” switch labeled as such. A Google Support page explicitly states that “Turning off ‘AI Overviews and more’ in Search Labs will not disable all AI Overviews in Search” and goes on to explain that “AI Overviews are part of Google Search like other features, such as knowledge panels, and can’t be turned off.” This official stance positions AI Overviews as a fundamental component of the modern Google Search experience, integrated into the core product rather than an optional enhancement. Google’s refusal to provide an official disable option reflects the company’s strategic commitment to making AI a central part of its search future, particularly as it competes with alternative search engines and AI-powered chatbots like ChatGPT.
Google’s official guidance does acknowledge that users have access to what is called the “Web” filter or the “More” menu where they can select “Web” view to see search results without AI Overviews. However, this is technically not disabling the feature so much as selecting a different view mode—users must actively click this filter for each search to avoid seeing AI Overviews. The Web filter strips results back to traditional blue links only, removing not just AI Overviews but also other search features like knowledge panels, featured snippets, video previews, and image results. While this provides one legitimate workaround, it requires users to take an active step for every single search query, making it impractical for many users who want a more permanent solution. Furthermore, Google has deliberately hidden this option within menus, requiring users to click “More” to access it, rather than displaying it prominently where users might naturally discover it.
Desktop Methods for Disabling or Avoiding AI Overviews
For desktop users employing Google Chrome or other Chromium-based browsers, several effective methods exist to avoid AI Overviews on a persistent basis, with the most reliable approach involving customizing the browser’s search engine settings rather than relying on extensions that are vulnerable to being broken by Google’s backend changes. The fundamental technique involves adding the URL parameter “&udm=14” to Google search queries, which directs the search engine to return only results from the “Web” tab, effectively bypassing AI Overviews entirely. This parameter, discovered and documented by users and technology journalists, works because Google uses internal codes to designate different result views and modes, with UDM=14 specifically targeting the traditional web results view. Users can implement this in two ways: manually editing each search query by adding the parameter to the URL in their address bar, or more practically, creating a custom search engine in their browser settings that automatically applies this parameter to every search they perform from the address bar.
To set up a custom Google search engine in Chrome that persistently applies the UDM=14 parameter, users should follow a straightforward process that takes approximately five minutes to complete. The first step involves opening Chrome and typing “chrome://settings/searchEngines” directly into the address bar, which navigates the user to the search engine customization panel. From this panel, users should select “Search Engine” followed by “Manage search engines and site search,” scrolling down to locate the “Site Search” section. Next to “Site Search,” users click the “Add” button to create a new custom search engine entry. In the dialog box that appears, users fill in three fields: the name can be anything descriptive such as “AI Free Web” or “Google Web,” the shortcut might be “@web” or a similar abbreviation for quick access, and critically, the URL field should contain the string “{google:baseURL}search?q=%s&udm=14.” After clicking “Add,” the new search engine appears in the list, and users must click the three-dot menu next to their new entry and select “Make Default” to ensure all searches from the address bar use this custom engine.
Once this custom search engine is established as the default, users will no longer see AI Overviews when searching from the address bar or from the search box on a new tab page. The results will display only traditional blue links without any of the additional visual features that Google has added to search over the past decade, including knowledge panels, featured snippets, image previews, video results, and shopping panels. This “clean” search experience approximates what Google search looked like in the early 2000s, before the company began layering additional features onto its results pages. Some power users and researchers have found this trade-off worthwhile, appreciating the reduction in visual clutter and the focus on primary sources, though others note that losing image, video, and other specialized result types can be limiting for certain types of searches.
An important limitation of this desktop method is that it only applies to searches initiated from the browser’s address bar or the search box on the new tab page. If users navigate directly to google.com in their browser, they will still see the full Google search experience with AI Overviews. This means users who have a bookmark to Google’s homepage or who type “google.com” directly into the browser will bypass their custom search settings. For users who want to ensure they never see google.com’s full interface, this might be a minor inconvenience, though it can be mitigated by making the custom search engine the default and avoiding direct visits to the Google domain.
For desktop users who prefer not to modify browser settings, browser extensions offer an alternative approach, though with significant caveats about long-term reliability. Extensions like “Hide Google AI Overviews,” “Bye Bye Google AI,” and similar tools use CSS code to visually hide the AI Overview section when it appears on the results page. The “Hide Google AI Overviews” extension, available on the Chrome Web Store and Firefox Add-ons, has accumulated over 200,000 users and maintains a 4.2-star rating. The advantage of these extensions is simplicity—users simply install them and the AI Overviews automatically disappear from view without any configuration needed. However, browser extensions are fragile solutions vulnerable to breaking whenever Google updates its Search page’s HTML structure or CSS classes. Google’s frequently released updates to its user interface have historically caused numerous search-related extensions to malfunction, requiring developers to publish fixes. This means extensions that work perfectly today might suddenly fail tomorrow, requiring users to either seek updates or reinstall them.
Another desktop approach involves using the advanced ad-blocking extension uBlock Origin, combined with custom CSS filter rules that target and hide the specific HTML elements Google uses for AI Overviews. Users of uBlock Origin can right-click directly on the AI Overview element and select “Block Element” to visually hide it, or more technically savvy users can add manual CSS filters to uBlock Origin’s “My Filters” tab. The filter “google.com##div[jsname=”yDeuDf”]” or variations thereof have been documented to effectively hide AI Overview sections, though the specific CSS selectors may change when Google updates its page structure. Like browser extensions, this approach requires the filter rules to be updated whenever Google changes its underlying code, but the open-source nature of uBlock Origin and the large community of users means that fixes are typically documented and shared relatively quickly.
Mobile Methods for Disabling or Avoiding AI Overviews
Disabling AI Overviews on mobile devices presents significantly greater challenges than on desktop, as mobile browsers typically do not support custom search engine configuration or advanced browser extensions in the same manner as desktop browsers. Google Chrome on mobile allows users to select a custom search engine, but the interface does not permit users to input the necessary “&udm=14” parameter manually, making the workaround used on desktop directly inaccessible to mobile users. However, a community-driven workaround has emerged in the form of websites like tenbluelinks.org and 10blinks.org, which function as helper sites specifically designed to facilitate the addition of the “Google Web” search engine to mobile browsers.
The process for mobile users employing this workaround involves first opening Chrome on their mobile device and visiting the helper website from their phone’s browser. Once on the helper site, users should open a new tab and perform any search on google.com—this initial search step is necessary to establish Google as a recently used search engine. After performing this search, users tap the three-dot menu icon in the bottom right corner of Chrome (or the menu appropriate to their browser), navigate to Settings, then Search Engine, where they will now see “Google Web” listed in the “Recently Visited” section. By selecting “Google Web” from this recently visited section, the browser registers this as an available search engine option, and users can set it as their default if desired. Once configured, all subsequent searches from the mobile browser’s address bar will use this Google Web filter, eliminating AI Overviews from appearing in results.
An alternative approach for Firefox users on mobile involves manually configuring a custom search engine directly within Firefox’s settings, a process that does not require a helper website. To accomplish this on Firefox, users navigate to the three-dot menu, select Settings, then Search, followed by “Default Search Engine,” and ultimately “Add Search Engine.” In the resulting dialog, users enter a name like “AI-free Web,” and in the search string field, input “google.com/search?udm=14&q=%s,” then save the configuration. This method is more cumbersome than the Chrome approach but offers a more permanent solution for Firefox mobile users without relying on external websites.
Users should be aware of trade-offs associated with mobile workarounds. When using the “&udm=14” filter on mobile, local search results—such as restaurants, businesses, or services near the user’s location—may not function as expected, since the Web filter strips away the specialized local search results panel that Google normally displays. This can be particularly problematic for mobile users who frequently rely on Google to find nearby services, restaurants, or businesses while away from a desktop. Additionally, the mobile workarounds require users to take initial setup steps and may not persist if browsers are updated or reset, potentially requiring reconfiguration.
For iOS users, the options are more limited, as iOS apps and Safari have different capabilities than Android browsers. Safari on iOS does not natively allow custom search engine configuration in the same way that Chrome or Firefox do, though users can download third-party apps like “Customize Search Engine” from the iOS App Store to facilitate custom search engine setup. The process involves downloading the app, configuring the custom Google search URL within the app, then enabling the app in Safari’s extensions settings. This approach is considerably more complex than the Android equivalents and requires downloading and managing additional applications specifically for this purpose.
Browser Extensions and Third-Party Tools for Removing AI Content
The market for browser extensions designed to hide or disable AI content across various Google products and search engines has grown substantially, reflecting the clear user demand for such tools. Beyond the specific “Hide Google AI Overviews” extension mentioned previously, several other notable tools have emerged to address different aspects of Google’s AI integration. The “Bye Bye Google AI” extension provides configurable options allowing users to selectively hide not just AI Overviews but also other elements they may find undesirable, including sponsored links, shopping blocks, video results, “People Also Ask” sections, and feature snippets. This granular control appeals to users who want to customize exactly which elements appear on their search results page rather than having a one-size-fits-all solution. The extension works by applying CSS rules to hide selected elements, and after installation, users must click the extension icon and select “Save” to activate their preferences, after which the hiding occurs automatically on subsequent searches.
The “Disable AI” extension takes a different approach by attempting to actually prevent AI features from running in the background rather than simply hiding them visually. This extension is designed to block Google’s AI Overview, DuckDuckGo’s AI Assist, Ecosia’s AI Overview, Brave Search’s Answer with AI, and Qwant’s AI Flash Answer, with the explicit goal of reducing the computational resources and water usage consumed by AI systems running in the background. The philosophical distinction is important: while most extensions merely hide AI results that have already been generated by the servers, the “Disable AI” extension aims to prevent the AI request from being made in the first place, thereby reducing energy consumption. This appeals to users concerned about the environmental implications of widespread AI deployment and those who prefer that their searches not contribute to server load for AI systems.
For Chrome specifically, an additional approach involves using Chrome flags—experimental settings accessible via the “chrome://flags” URL—to disable various AI-related features embedded throughout the browser itself, rather than just the search functionality. Users can search for “AI mode” within the flags page and disable the “AI Mode Omnibox Entrypoint” flag to remove the AI Mode button from the address bar. Similarly, disabling flags such as “AI Entrypoint Disabled on User Input” and “Omnibox Allow AI Mode Matches” can prevent AI Mode from appearing in other contexts. Additionally, users can navigate to “chrome://settings/ai/historySearch” to disable Google’s AI-powered search history feature. A significant limitation of using Chrome flags is that they constitute experimental settings that Google may modify, remove, or reset with browser updates, making them an unreliable long-term solution. Google updates Chrome every few days, and these updates sometimes reset flag settings back to default values, requiring users to repeatedly disable the same flags.
The “U14 Extension” has been specifically designed to work alongside the standard Google search to ensure that the “&udm=14” parameter is always applied, making it easier for users to maintain their preference for non-AI search results even if they accidentally navigate to google.com directly. This extension essentially automates the custom search engine approach by intercepting Google searches and automatically redirecting them through the Web filter, reducing the likelihood that users will accidentally encounter the full AI-enhanced search experience.
It is crucial for users to understand that all browser extensions designed to modify Google Search results are inherently fragile and vulnerable to breaking whenever Google makes changes to the underlying code of its search results pages. Google’s development teams are constantly updating the HTML structure, CSS classes, and JavaScript that power search results, and when these updates occur, extensions that rely on targeting specific elements by their names or classes may fail. This is not a malicious act by Google, but rather a natural consequence of ongoing product development, yet it means that users who rely on extensions must be prepared to seek updates or switch to alternative approaches if an extension suddenly stops working.
Alternative Search Engines as Comprehensive Solutions
For users frustrated with Google’s aggressive AI integration and unwilling to engage in perpetual workarounds, switching to alternative search engines represents a comprehensive solution that eliminates the need for ongoing technical tweaks and configuration. The search engine landscape has diversified considerably in recent years, with numerous alternatives offering different values propositions ranging from privacy-first approaches to AI-enhanced search experiences that users may find more reliable than Google’s offerings. Understanding the options available and their respective strengths and limitations is essential for users evaluating whether to abandon Google Search entirely.
Privacy-focused search engines like DuckDuckGo and Brave Search have gained significant user bases by positioning themselves as alternatives to Google’s tracking-based model. DuckDuckGo, one of the earliest and most established privacy-focused search engines, does not track user search history, does not create user profiles, and does not serve personalized advertisements based on search behavior. The search engine sources results from multiple providers, including Bing and other data sources, and explicitly does not employ AI-generated summaries or overviews in its standard search results, making it an ideal choice for users seeking to avoid AI summaries entirely. Brave Search represents a newer entry to the privacy-focused market, having developed its own independent search index starting in 2023 rather than relying on Google or Bing’s infrastructure. Like DuckDuckGo, Brave Search does not track users and does not use personal search history to personalize results. Both of these search engines are available as smartphone apps, allowing mobile users to avoid AI Overviews through app-based search without needing to configure custom search engines or visit helper websites.
For users who appreciate AI-enhanced search but want more reliable and accurate results than Google’s AI Overviews, Perplexity AI represents a compelling alternative that has become increasingly popular since its launch in 2022. Perplexity functions as a conversational AI search engine that combines chatbot-like interaction with real-time web searches, presenting information in a more structured and transparent format than Google’s approach. When users search on Perplexity, the system provides a response with sources clearly listed and highlighted, allowing users to easily verify the information and understand where it came from. Perplexity’s interface includes a “Sources” tab showing all citations, a “Steps” tab revealing the reasoning process the AI followed, and a “Related” section suggesting follow-up questions, all of which promote transparency and allow users to dig deeper into topics. The platform offers a free tier with basic capabilities and a Pro subscription ($20/month) providing access to more advanced models and higher query limits. Importantly, Perplexity has cultivated a reputation for relatively high accuracy compared to other AI systems, with users reporting fewer hallucinations and more reliable citations.
ChatGPT Search, OpenAI’s search-integrated version of ChatGPT officially launched in October 2024, offers another alternative for users seeking AI-enhanced search with reliable citations and conversational capabilities. Unlike Google’s AI Overviews, which appear alongside traditional search results, ChatGPT Search fully replaces the traditional search experience with a conversational AI interface, similar to interacting with ChatGPT but with real-time web data access. This approach appeals to users already familiar with ChatGPT who want the benefits of web search integration without switching platforms. The service is now available to all users after initially launching to paid subscribers and enterprise customers.
Kagi represents another alternative worth consideration, particularly for users willing to pay for search services. Kagi is an ad-free, privacy-focused search engine that maintains its own independent search index and offers powerful customization features through its “Lenses” system, which allows users to filter results to prioritize forums, academic papers, indie blogs, or other content types. While Kagi’s $10/month subscription may deter casual users, the platform’s emphasis on quality, independence, and customization appeals to power users and professionals who conduct extensive research. Kagi explicitly does not use AI-generated summaries in its standard search results, providing pure search results without AI intermediation.
For users seeking specialized or niche alternatives, Dogpile functions as a metasearch engine aggregating results from multiple sources including Google, Bing, Yandex, and Yahoo, allowing users to compare results across multiple search engines without relying solely on any single platform. WolframAlpha serves users with computational and data-heavy queries, offering structured answers to mathematical, scientific, and technical questions. These specialized alternatives cannot fully replace Google for all search needs but can supplement general search with targeted results for specific query types.
AI Mode Versus AI Overviews: Understanding the Broader Landscape
To comprehensively understand the current state of Google’s AI-driven search landscape, it is essential to distinguish between AI Overviews and Google’s newer AI Mode, two related but fundamentally different features that represent different philosophies for integrating artificial intelligence into search. Confusion between these two features is common among users, as both utilize Google’s Gemini AI model and both provide AI-generated content, yet they function in substantially different ways with vastly different implications for publishers, content creators, and the structure of information on the web.
AI Overviews, as discussed extensively throughout this report, present synthesized summaries at the top of the traditional search results page, with traditional blue links remaining visible below the AI-generated content. Users still see and can interact with the full Google Search experience, including knowledge panels, image results, video results, shopping panels, and other specialized search features. The AI Overviews feature represents an enhancement or augmentation of traditional search rather than a replacement of it. The feature breaks queries into component questions through query fan-out to provide more comprehensive answers, but the responses are fundamentally static summaries—users cannot engage in back-and-forth conversation or ask follow-up questions within the AI Overview itself.
AI Mode, by contrast, represents a far more dramatic departure from traditional search, essentially replacing the entire Google Search experience with a conversational AI interface similar to ChatGPT. When users switch to AI Mode—available through a dedicated tab or button in Google Search—they no longer see the traditional 10 blue links. Instead, the interface becomes a conversational chatbot where users can ask questions, receive AI-generated answers, and engage in multi-turn conversations with follow-up questions and clarifications. AI Mode uses a more sophisticated version of the Gemini model trained specifically for reasoning and exploration, and it can access real-time data including stock market information, restaurant availability, and Google’s Shopping Graph. The experience more closely mirrors using ChatGPT with web search capabilities than using traditional Google Search.
The implications of AI Mode for publishers and content creators are far more severe than even those associated with AI Overviews. Early data suggests that traffic declines from AI Mode could be substantially more dramatic than the 30-70% range reported for AI Overviews, potentially making many publishing businesses unviable. Unlike AI Overviews, where links are at least present on the page, AI Mode displays only 1-3 sources in its initial response, with users needing to click “Show all” to view additional sources. This represents the information compression that publishers fear: users receive their information directly from AI without any incentive to visit the original source websites. Industry experts warn that if Google makes AI Mode the default search experience—which Google’s CEO Sundar Pichai suggested the company is considering—it could have “devastating impact on the internet” by severely reducing website traffic and disincentivizing the creation of original content that AI systems then use without properly compensating creators.
The Broader Consequences: Publisher Impacts and Industry Response
The rollout of AI Overviews has generated significant backlash from independent publishers, news organizations, and content creators who contend that the feature is directly responsible for reduced website traffic and declining revenue. Multiple research reports have documented substantial traffic declines correlated with AI Overviews’ rollout, with BrightEdge research showing that AI Overviews have caused impressions to rise 49% across the web while clicks have simultaneously fallen 30%. This apparent paradox—more impressions but fewer clicks—reflects the core problem publishers face: Google is showing their content in search results more frequently, but users are clicking through to their websites less often because they are obtaining their information from the AI summary instead. Some publishers report traffic losses in the 20-60% range depending on their content category and search terms that trigger AI Overviews.
The Independent Publishers Alliance filed a formal complaint with the European Commission in June 2025, alleging that Google is abusing its market dominance in search through AI Overviews that appear without publisher opt-out capability and that drive traffic away from publishers’ websites. The complaint argues that publishers cannot opt out of having their content used for AI summaries without entirely losing their ability to appear in Google Search results, presenting an impossible choice between losing SEO visibility or having their content used to populate AI summaries that reduce traffic to their sites. This situation has led to litigation, with Chegg, a student education platform, suing Alphabet in February 2025 over AI Overviews, claiming the feature violated antitrust law by encouraging students to rely on “low-quality, unverified AI summaries” rather than Chegg’s offerings. Similarly, Penske Media Corporation, publisher of Rolling Stone and The Hollywood Reporter, sued Google in September 2025, alleging that AI Overviews illegally regurgitate content from their websites without fair compensation and appear in positions that reduce incentive for users to visit the original sources.
Research on user behavior provides empirical support for publishers’ concerns about traffic declines. A Pew Research Center analysis of browsing data from March 2025 found that Google users who encountered AI Overviews were significantly less likely to click on traditional search result links compared to users who did not see AI summaries. Specifically, users encountering AI Overviews clicked on traditional search result links in only 8% of visits, compared with 15% of visits for those who did not see AI summaries. Even more striking, users very rarely clicked on links cited within the AI Overviews themselves, with this occurring in just 1% of all visits to pages with AI summaries. Additionally, users were substantially more likely to end their entire search session after viewing a page with an AI summary without clicking anything, occurring in 26% of visits to pages with AI summaries compared to 16% for pages without summaries.
Accuracy, Quality, and Reliability Concerns
While AI Overviews have improved since their introduction in early 2024, persistent accuracy and quality concerns continue to plague the feature, raising questions about whether AI-generated summaries should be entrusted with the role of answer provider in search results. The early months of 2024 saw numerous embarrassing and sometimes dangerous examples of AI Overviews providing false, misleading, or potentially harmful information, from the infamous “use glue on pizza” suggestion to recommendations for health conditions that contradicted established medical practice. A Business Insider analysis in late 2025 acknowledged that while AI Overviews have improved throughout the year, the feature continues to struggle with accuracy, and the publication noted that earlier concerns about quality have not entirely disappeared.
The Guardian conducted an investigation of AI Overviews for health-related queries, involving health organizations and experts reviewing the AI-generated summaries and raising concerns about inaccuracies that could pose genuine risks to user health. Health experts pointed out that even when AI Overviews link to reputable sources, the summaries themselves can override user trust by presenting incorrect guidance confidently and without appropriate qualification. One example involved an AI summary that appeared to provide appropriate nutritional advice but, when examined closely, could lead to inadequate caloric intake for cancer patients, potentially compromising their ability to tolerate chemotherapy or life-saving surgery. Google disputed some of the Guardian’s examples, claiming they were “incomplete screenshots,” but the investigation highlighted a fundamental tension: Google holds publishers to rigorous standards for health content accuracy, yet has not been held to the same standard for its own AI-generated summaries.
AI’s tendency toward hallucination—generating plausible-sounding but false information—remains a significant limitation. Testing has shown that when users search for deliberately nonsensical phrases such as “you can’t lick a badger twice” or “you can’t fit a duck in a pencil,” early versions of AI Overviews would confidently generate explanations of these non-existent idioms, treating them as real language concepts. While Google has improved its handling of such false-premise searches, the system can still struggle with edge cases and unusual queries that fall outside the training data’s common patterns. The underlying problem is that large language models operate by finding statistical patterns in text and generating outputs that are statistically likely given those patterns, not by genuinely understanding truth in the way humans do. When web content contains false information, myths, conspiracy theories, or fringe views presented with confidence, AI systems have difficulty distinguishing them from reliable information.
Taking Control of Your Google Search
As Google continues to aggressively integrate artificial intelligence into its search product and other services, users face a complex landscape of options for managing their search experience and protecting their preferences. The company has made clear through its actions and official statements that AI Overviews and AI Mode are fundamental to its vision for the future of search, and that no simple official “off” switch exists for those who wish to opt out. Yet the persistence of user demand for non-AI search, combined with the documented accuracy concerns, publisher impacts, and fundamental philosophical disagreements about how information should be presented to users, suggests that the tension between Google’s AI ambitions and user preferences will remain unresolved for the foreseeable future.
For desktop users, the most reliable solution remains the customization of browser search engine settings to persistently apply Google’s “&udm=14” parameter, which directs searches to the traditional Web view without AI Overviews. This approach avoids vulnerability to extension breaks and requires no ongoing maintenance beyond the initial setup. Mobile users must engage in more complex workarounds involving helper websites or manual configuration, reflecting the more limited customization options available on smartphone platforms. For users unwilling to employ these workarounds, switching to alternative search engines—whether privacy-focused options like DuckDuckGo and Brave, AI-enhanced alternatives like Perplexity and ChatGPT Search, or premium options like Kagi—provides comprehensive solutions that eliminate AI Overviews entirely by removing dependence on Google’s platform.
The broader landscape suggests that Google’s dominance in search, while formidable, is not unquestionable, and the company’s aggressive AI integration strategy has created an opening for competitors offering different values propositions. As AI Mode potentially becomes more prominent in 2026 and beyond, the stakes for this competitive landscape will increase, with publishers and content creators potentially facing even more severe traffic impacts if Google makes AI Mode the default search experience. The ultimate resolution of this tension will likely depend on regulatory action, user behavior changes, and Google’s willingness to adapt its product roadmap in response to documented harms and user preferences.
For the individual user seeking to disable or avoid AI Overviews, the message is clear: while official options remain limited, multiple practical workarounds exist that are effective if not always convenient. Whether through custom search engine configuration, browser extensions, alternative search engines, or some combination thereof, users retain agency in determining their search experience, even as that agency requires increasing technical sophistication and active choice-making to maintain in the face of Google’s push toward an AI-first paradigm.
