The Catch Behind Google AI U-turn in Schools

Google AI in schools is about to change. The company has announced that its artificial intelligence tools will be switched off by default for schoolchildren in the United States. At first glance, the decision sounds like a significant victory for parents and educators concerned about AI in schools.

However, there is a catch. Schools that want pupils to use Google Search will still have to accept AI features built into the search experience.

The 3 November announcement, highlighted by RTÉ News, raises an uncomfortable question. If a school decides that generative AI is not appropriate for its pupils, why should it have to give up access to ordinary search tools? This debate is about more than one company’s settings. It concerns independent learning, child safety and who gets to decide what technology belongs in the classroom.

Google AI in schools, what has changed?

In its official announcement, Google says access to Gemini, Gemini Notebook, Gemini in Classroom and Google Search will be off for K–12 student accounts until school administrators explicitly enable the products. Schools can make different decisions about individual services. Google says it listened to teachers, researchers and policymakers who wanted more control.

Importantly, this is a planned change, not proof that all American schools have already switched these products off. Google said it would share further details ahead of the rollout. Education Week reported that the new controls are expected to begin in early 2027, while a pilot of Guided Learning in Search is due to start in selected US schools in November 2026.

However, the crucial limitation remains. Schools cannot choose an ordinary version of Google Search that removes AI Overviews and AI Mode through these controls. According to Education Week, Google says Search and AI are deeply integrated. Therefore, if a school prohibits the integrated AI features, students lose Google Search access through that school account.

A choice between AI search and no Google Search

For example, Google’s decision gives administrators an important new default setting. Nevertheless, switching off a product is different from providing a genuine choice about the features within it. A school may value the ability to research primary sources, compare competing claims and practise finding reliable information online. Those skills do not require a chatbot to generate answers before pupils examine the evidence.

Yvette Renteria, chief programme officer at Common Sense Media and a former school administrator, warned Education Week about the educational consequences. She argued that pupils need opportunities to find and evaluate sources. She also questioned whether Google’s new Guided Learning tools had received sufficient independent scrutiny before expansion.

Former Federal Trade Commission chair Lina Khan criticised Google’s use of its position in search. Senator Richard Blumenthal also challenged the decision, while education leaders raised concerns about the practical burden on schools. These reactions reflect a broader issue, the company selling the technology should not be the only party defining what counts as an acceptable educational choice. Read the full Education Week coverage.

Independent tests raise serious questions

Moreover, the controversy becomes more significant when considered alongside independent research. In July 2026, the Common Sense Media Youth AI Safety Institute published an assessment of Google’s AI Overviews and AI Mode. Researchers conducted more than 2,600 searches using accounts configured for children aged 11 and 15, with SafeSearch enabled. The institute gave the products its lowest assessment, Unacceptable Risk.

Importantly, its findings concern safety and reliability under controlled test conditions. They do not show that every child using Google Search will experience harm. However, they raise clear questions about making AI-generated answers a routine feature of children’s learning.

What the researchers found

• 180 out of 180: AI Mode completed every homework assignment submitted during the study, rather than consistently encouraging pupils to work through the task themselves.
• 43%: Repeated history questions returned materially different answers in this proportion of tests.
• 29%: Of 2,100 citations examined, this proportion came from user-generated sources such as forums and social platforms.
• 58%: AI Overviews recognised clear warning signs of distress in this proportion of the researchers’ crisis-related test searches.

These figures come from the institute’s published explanation of its testing. The 58% measure refers to recognising warning signs in simulated crisis searches, not to the proportion of children who experience a mental health crisis. Similarly, the homework figure measures responses to a defined set of assignments, rather than proving that all AI-assisted homework prevents learning.

Can Guided Learning solve the problem?

Meanwhile, Google is developing Guided Learning features intended to prompt students through problems instead of immediately handing them a finished answer. Its classroom approach could help teachers structure supported practice, and the company says educational research informs the design. That deserves fair consideration, particularly where teachers remain involved and learning outcomes can be measured.

Nevertheless, a product promising better learning is not the same as independent evidence that it is safe and effective for children of different ages. New functionality should be evaluated for accuracy, age appropriateness, privacy, emotional safety and educational outcomes. Evaluation should also consider whether a pupil can reach traditional sources without first being steered towards an AI-generated response.

US lawmakers demand answers

On 21 October 2026, US Senators Richard Blumenthal and Josh Hawley sent a bipartisan letter to Google CEO Sundar Pichai. They requested information about academic dependence, children’s wellbeing, AI safety research and the company’s integration of AI into school services. Their concerns followed reporting on pupils using chatbots for academic work and sensitive personal conversations.

The lawmakers’ allegations should not be treated as independently established findings about every Google product. Yet the questions deserve answers. A child may start by asking an assistant for help with mathematics and later turn to the same system for reassurance about friendships, anxiety or other personal problems. Educational settings must account for both patterns of use.

Why Irish schools should pay attention

Nevertheless, the American change does not automatically establish what controls will apply to Irish school accounts. Google has not confirmed identical timing or implementation for every country. Nonetheless, the debate is relevant to Ireland because digital classroom platforms increasingly combine familiar learning services with generative AI.

Ireland’s Department of Education published national guidance on AI in schools in October 2025. It subsequently announced an external advisory taskforce in April 2026 to examine AI in teaching, learning and assessment. These initiatives create an opportunity to ask whether schools can decline embedded AI without losing basic educational functionality.

Furthermore, the international context also matters. UNESCO’s guidance on generative AI in education supports age-appropriate approaches and human-centred decision making. Irish schools should be able to examine privacy, training data, accuracy, learning outcomes and the consequences of reliance on automated answers before adopting any product.

What parents and school leaders can do

• Ask which AI features pupils can access through school Google accounts and whether the controls apply differently by age or class.
• Check whether teachers can require conventional research using original sources and whether pupils are taught to verify generated claims.
• Seek details of independent safety evaluations before enabling classroom AI tools, particularly for primary school children.
• Discuss how pupils should respond when an AI tool gives confident but inaccurate information or invites sensitive personal disclosures.
• Review CODA’s Google Search safety settings guide and Gemini safety settings guide while remembering that account controls vary by country and product.

Schools should protect time for reading, writing, conversation, play and learning without unnecessary digital intervention. Equally, when technology genuinely supports a lesson, it should remain a tool chosen by educators rather than a condition imposed by a platform provider.

Children’s learning must come first

At Children of the Digital Age, we support a smartphone-free childhood, reduced recreational screen time and keeping social media out of national school childhood. We also believe children need age-appropriate digital literacy. That includes recognising when AI can help, when it can mislead and when independent thought is essential.

Google’s decision to turn products off by default acknowledges that schools need more say. However, genuine choice means being able to keep essential search functionality without accepting features a school considers unsuitable. Parents and educators should not have to choose between access to information and exposure to generative AI.

Children’s education should be shaped by evidence, independent safety testing and the needs of learners, not by the product design decisions of a technology company.

References

• Google: Designing responsible AI for learning (3 November 2026)
• Education Week: Google’s new rules hand schools hard choices (3 November 2026)
• Common Sense Media: Google AI Search risk assessment (15 July 2026)
• US Senate: Blumenthal and Hawley letter to Google (21 October 2026)
• Department of Education: Guidance on AI in schools (21 October 2025)
• Department of Education: AI advisory taskforce (7 April 2026)
• UNESCO: Guidance for generative AI in education and research

Further Reading

• Google’s Family Link App for IOS and Android
• Google Family Link Safety Settings
• Parental Consent Is Not Child Protection
• Do Parents Know Children Are Turning to AI?

About Children of the Digital Age

Children of the Digital Age provides online safety education and training for schools, parents and organisations. Our guides support practical conversations about online behaviour and the choices people make. However, education and app settings do not replace safeguarding or professional support.

Send a message

Contact Children of the Digital Age to enquire about online safety education, parent sessions or training for your school or organisation.