Common Sense Media Labels Google AI Search an Unacceptable Risk for Children

The “Unacceptable Risk” Assessment: How Common Sense Media Tested Google Search

In a report released in mid-July 2026, the Youth AI Safety Institute at Axios concluded that Google’s generative AI search features pose an unacceptable risk to minors. The nonprofit, which evaluates technology for educational and safety standards, conducted more than 2,600 searches using accounts configured for 11- and 15-year-olds. According to the findings, these tools failed to adhere to seven of the organization’s eight core AI safety principles.

The evaluation focused on two specific features: Los Angeles Times, which appear at the top of results, and AI Mode, a conversational tab. Researchers found that the tools frequently treated forums and social media posts as equivalent to peer-reviewed research or medical institutions. More critically, the system failed to consistently identify or redirect users exhibiting signs of suicidal ideation, psychosis, or disordered eating.

Failure to Detect Crisis and Academic Integrity Concerns

The report highlighted specific instances where the AI provided dangerous or inappropriate responses to test accounts. When researchers simulated a user expressing symptoms of disordered eating, the AI described purging as completely normal. In other cases, the tools recommended crisis hotlines that had been disconnected since 2023. Researchers also noted that the AI provided step-by-step instructions for creating deepfakes, which could be leveraged for bullying.

Beyond safety, the report raised concerns regarding academic integrity in K-12 settings. AI Mode successfully completed 100 percent of the homework assignments researchers submitted. Unlike standalone chatbot products, these features are integrated directly into the search experience on school-issued devices, such as Chromebooks, and cannot be toggled off by administrators or parents.

“There is no reason why a 10-year-old, 13-year-old, or even 15-year-old should see an AI Overview or have AI Mode. When teaching AI literacy, the first step is to be intentional. These modes are the opposite. You don’t go to a search engine to get a chatbot answer, you go to search for real things.”

Amanda Bickerstaff, AI for Education

Google’s Response and the Debate Over Product Safety

Google has pushed back against the report’s methodology. Google spokesperson Davis Thompson stated that the study utilized a narrow set of ambiguous and contrived queries that do not accurately represent how users typically interact with Search. The company maintains that its AI features include robust safety guardrails and provide useful tools for exploration and learning.

While acknowledging that the AI features cannot be disabled individually, Google noted that parents and administrators have the option to turn off Search entirely for child-managed accounts. However, advocates argue that this is an impractical solution for students who rely on Google tools for their education. As Amanda Bickerstaff of AI for Education noted, the ubiquity of Google in classrooms makes blocking the entire search engine an unrealistic request for most schools.

Stakes for Schools and Regulatory Outlook

The findings arrive at a time of increased scrutiny regarding the role of AI in youth digital environments. With 75% of U.S. teens and tweens already using AI-generated search results, experts are concerned about the long-term impact on information literacy. Robbie Torney, senior director of AI at Common Sense Media, emphasized that because Google Search is a default tool on millions of student devices, any failure in its safety design carries significant, population-scale consequences.

For now, the debate remains centered on whether the convenience of generative search outweighs the risks posed to a generation of students currently navigating these tools in their daily coursework.