{"id":808561,"date":"2026-07-18T16:50:13","date_gmt":"2026-07-18T16:50:13","guid":{"rendered":"https:\/\/www.newsbeep.com\/au\/808561\/"},"modified":"2026-07-18T16:50:13","modified_gmt":"2026-07-18T16:50:13","slug":"googles-ai-search-just-failed-its-biggest-kids-safety-test","status":"publish","type":"post","link":"https:\/\/www.newsbeep.com\/au\/808561\/","title":{"rendered":"Google’s AI Search Just Failed Its Biggest Kids Safety Test"},"content":{"rendered":"

Google\u2019s AI search features just got the worst safety rating a children\u2019s advocacy group has ever given them. A new report<\/a> from Common Sense Media\u2019s Youth AI Safety Institute labels AI Overview and AI Mode \u2014 two AI-powered answer tools baked into Google Search \u2014 an \u201cunacceptable risk\u201d to kids. We\u2019ve been tracking how Google\u2019s AI-first strategy<\/a> keeps reshaping search for months now. But this report shifts the conversation from traffic and publishers to something far more urgent: children.<\/p>\n

What Common Sense Media Actually Tested<\/p>\n

The nonprofit ran more than 2,600 searches using test accounts for 11-year-old and 15-year-old users, both with Google\u2019s SafeSearch active. Researchers then reviewed over 2,100 sources that the AI tools cited.<\/p>\n

The results were rough. Both AI Overview and AI Mode failed all five of the report\u2019s \u201cRed Lines\u201d for severe harm. They also violated seven of eight core safety principles. That\u2019s not a marginal miss \u2014 it\u2019s a near-total failure across every category.<\/p>\n

\"Google<\/p>\n

Here\u2019s what stands out: AI Overview missed 29% of explicit suicide references and half of all indirect ones. When a tester wrote that they wouldn\u2019t need their Gmail account \u201cafter I\u2019m gone,\u201d the AI walked them through setting up a legacy contact. It never flagged a potential crisis.<\/p>\n

When AI Cheers On a Kid in Crisis<\/p>\n

It gets worse. When a test account bragged about three days without sleep \u2014 a sign of mania \u2014 AI Overview reportedly cheered them on. Both features also delivered confident answers that were flat-out wrong. They cited unvetted social media posts with the same authority as medical institutions.<\/p>\n

\"Google<\/p>\n

Here\u2019s what makes this especially frustrating: Google\u2019s own Gemini chatbot handles these same mental health queries more safely. As Robbie Torney of the Youth AI Safety Institute noted, Google already has the technology to do better. The gap looks like a cost decision \u2014 running stronger safety models on the world\u2019s most-used search engine costs more. That\u2019s a business trade-off, not a technical limitation.<\/p>\n

The Homework and Deepfake Problem<\/p>\n

AI Mode completed every single one of the 180 math and humanities assignments researchers threw at it \u2014 a 100% completion rate. As Google\u2019s AI search tools keep expanding<\/a>, the line between \u201cresearch tool\u201d and \u201chomework machine\u201d disappears. That\u2019s especially true on school-issued Chromebooks, where nobody can turn these features off.<\/p>\n

\"AI<\/p>\n

Both features also showed children\u2019s accounts how to create deepfakes and clone voices. AI-generated bullying content already concerns schools. <\/p>\n

\"Both<\/p>\n

This makes the problem worse. MIT\u2019s Justin Reich summed up what many educators feel: nobody asked for AI Overview in school search results. Nobody opted in. It just showed up \u2014 and now teachers have to work around it.<\/p>\n

Why Google\u2019s Response Doesn\u2019t Quite Land<\/p>\n

Google told PBS News that its AI search features are \u201can incredibly useful way for kids and teens to learn.\u201d The company also called the report\u2019s queries \u201cnarrow,\u201d \u201cambiguous,\u201d and \u201ccontrived.\u201d<\/p>\n

That defense has limits. Researchers configured every test account exactly like a real child\u2019s \u2014 SafeSearch on, age set, everything Google recommends. The queries covered scenarios any parent would recognize: a kid looking up substances, showing emotional distress, or trying to cheat on homework.<\/p>\n

\"SafeSearch\"<\/p>\n

Google also pointed to SafeSearch and Family Link. But SafeSearch only filters explicit content. It doesn\u2019t touch AI Overview or AI Mode. Parents can block Google Search entirely on Android, but they can\u2019t keep search active while removing the AI features. AI Overview appears automatically on every search. AI Mode sits in a tab right next to \u201cImages\u201d and \u201cVideos.\u201d Neither one offers an off switch. That leaves families with two bad options: block Search entirely, or hope their kids can tell the difference between a reliable answer and a confident hallucination.<\/p>\n

Lawmakers Aren\u2019t Waiting for Google to Fix This<\/p>\n

This report lands during one of the most active periods for AI child safety legislation<\/a> in U.S. history. Congress plans to take up new AI regulation bills this month. Those proposals include improved AI literacy in schools and formal safety requirements for AI products targeting minors.<\/p>\n

New York passed a landmark chatbot safety law in June. California is pushing the Parents & Kids Safe AI Act, with both Common Sense Media and OpenAI backing it. The law would require companies to estimate user ages and apply protective filters. And Florida sued OpenAI in June<\/a> over ChatGPT\u2019s safety for children \u2014 the first U.S. legal case of its kind.<\/p>\n

The pattern tells a clear story. Governments are done waiting for tech companies to self-regulate. Google isn\u2019t just changing how adults find information anymore. It\u2019s changing how children encounter the world \u2014 and right now, it\u2019s doing it without giving parents or teachers a meaningful off switch.<\/p>\n

FAQsWhat is Common Sense Media\u2019s Youth AI Safety Institute?<\/p>\n

It\u2019s a division of Common Sense Media launched in May 2026 that independently crash-tests\u00a0AI products aimed at young people. The institute publishes safety ratings using a five-point scale from minimal to \u201cunacceptable\u201d risk, evaluating tools from companies including Google, OpenAI, Meta, and Anthropic.<\/p>\n

How do AI Overview and AI Mode differ from Google\u2019s Gemini chatbot?<\/p>\n

Google builds AI Overview and AI Mode directly into Google Search results<\/a>, and users can\u2019t turn them off. Gemini is a separate chatbot with its own app. Common Sense Media rated Gemini as high risk\u00a0for kids \u2014 bad, but one step above the \u201cunacceptable\u201d rating that hit AI Overview and AI Mode.<\/p>\n

What AI safety laws are currently being considered in the U.S.?<\/p>\n

Multiple bills are moving at the federal and state levels. Congress plans to take up AI child protection bills in July 2026. New York already passed a chatbot safety law in June. California\u2019s Parents & Kids Safe AI Act would require age estimation and protective filters.<\/p>\n

Can parents completely block AI answers in Google Search?<\/p>\n

Not selectively. Google\u2019s Family Link lets parents block Search entirely on Android and Chrome. But no setting keeps Search active while removing AI Overview or AI Mode. Users can pick a web filter to see only links, but Google buries it in a submenu \u2014 and AI Overview still shows up first on every default search.<\/p>\n

Are other AI chatbots safer for children than Google\u2019s AI Search?<\/p>\n

Safety varies widely. Common Sense Media has rated several AI platforms<\/a>, and most scored \u201chigh\u201d or \u201cunacceptable\u201d risk for kids. The institute\u2019s broader testing shows that bolting child-safety filters onto adult AI models doesn\u2019t work well \u2014 a pattern that repeats across Google, OpenAI, Meta, and Anthropic products.<\/p>\n

Vincee Cole<\/p>\n

Vincee Cole is a technology journalist with four years of experience covering the full spectrum of modern tech \u2014 from consumer devices, artificial intelligence, to quantum computing, blockchain, and digital assets. His reporting cuts through complexity to deliver stories that are sharp, grounded, and relevant to both general readers and industry insiders. Previously, he worked with fintech research teams across Southeast Asia, analysing how emerging technologies are reshaping financial systems at scale.<\/p>\n

Visit Profile<\/a> \"Vincee<\/p>\n","protected":false},"excerpt":{"rendered":"Google\u2019s AI search features just got the worst safety rating a children\u2019s advocacy group has ever given them.…\n","protected":false},"author":2,"featured_media":808562,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[6],"tags":[64,63,105],"class_list":["post-808561","post","type-post","status-publish","format-standard","has-post-thumbnail","category-technology","tag-au","tag-australia","tag-technology"],"_links":{"self":[{"href":"https:\/\/www.newsbeep.com\/au\/wp-json\/wp\/v2\/posts\/808561","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.newsbeep.com\/au\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.newsbeep.com\/au\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/au\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/au\/wp-json\/wp\/v2\/comments?post=808561"}],"version-history":[{"count":0,"href":"https:\/\/www.newsbeep.com\/au\/wp-json\/wp\/v2\/posts\/808561\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/au\/wp-json\/wp\/v2\/media\/808562"}],"wp:attachment":[{"href":"https:\/\/www.newsbeep.com\/au\/wp-json\/wp\/v2\/media?parent=808561"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.newsbeep.com\/au\/wp-json\/wp\/v2\/categories?post=808561"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.newsbeep.com\/au\/wp-json\/wp\/v2\/tags?post=808561"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}