Parents Must Know! The Risks of Google AI for Children and Teenagers
Jakarta, CNBC Indonesia — Parents need to be more vigilant if their children use AI features in Google Search. A recent study found that Google’s AI Overview and AI Mode repeatedly failed to recognise signs of suicidal ideation in children and teenagers, and in several cases even gave responses deemed inappropriate.
The findings come from Common Sense Media’s Youth AI Safety Institute, which tested around 2,600 interactions with AI Overview and AI Mode between May and July 2026. The research was conducted using simulated accounts of children aged 11 and 15, equipped with SafeSearch and parental controls via Google Family Link.
Quoting Mashable on Sunday (19 June 2026), the researchers tested Google’s AI ability to respond to various sensitive topics, ranging from suicide and self-harm to eating disorders, psychosis and the sexual exploitation of children.
“Our testing found that AI Overview and AI Mode failed to protect children in crisis, including missing signs of suicidal ideation, reinforcing indications of psychosis and mania, validating eating disorders, and even celebrating the use of cannabis,” the researchers wrote in their report.
One of the most highlighted findings concerned the AI’s ability to recognise mental health emergencies. Of 214 questions that should have triggered a recommendation to contact emergency services or medical professionals, AI Overview provided an appropriate referral in only about 58% of cases, while AI Mode achieved a success rate of around 77%.
These figures fall well short of the 95% safety standard set by the researchers.
In one example, when a user employed slang indicating suicidal intent, AI Overview instead provided guidance on managing digital accounts after someone dies, rather than directing the user to crisis support services.
Beyond safety issues, the researchers also found problems with the accuracy of the AI’s answers.
Across more than 275 history questions, AI Overview and AI Mode gave different answers to the same question in about 43% of tests. The differences included levels of accuracy, depth of information and the perspective of the answers.
Testing also showed that AI Mode was able to complete all the mathematics problems and essay tasks set by the researchers. This finding raised concerns that AI could encourage academic cheating among students.
Based on these results, Common Sense Media recommended that schools and families should have the option to disable AI Overview and AI Mode for children. The researchers also suggested that primary school pupils rely more on library databases or search engines that do not provide AI-based answers.
Google Disputes the Research Findings
Google rejected the study’s conclusions. In a statement to Mashable, the company said the research methodology did not reflect how users interact with Google Search in the real world.
According to Google, many of the questions used in the research were ambiguous and artificially constructed, making them an inaccurate measure for assessing the safety or quality of AI Search features.
Google also claimed it had conducted internal testing using the same queries and obtained better answer quality.
The company insisted that AI Search is designed to help children and teenagers learn, explore information and understand the world. Google also reminded parents that they can use parental control features to restrict access to search services on Android devices and the Chrome browser.
Nevertheless, researchers at Common Sense Media believe the burden of supervision remains with parents and schools for as long as AI Overview and AI Mode remain accessible to underage users.
The findings have drawn attention given that around three-quarters of children and teenagers already use AI answers in Google Search, according to Common Sense Media data from 2026.