Google pushes back on report labeling its AI Search tools unsafe for

Google said its AI Overview and AI Mode features are safe for children, rejecting a Common Sense Media report that called them an “unacceptable risk” and urging that the study’s 2,600 test queries do not reflect real‑world use [1].

Google’s rebuttal to the safety report

The Youth AI Safety Institute, part of Common Sense Media, evaluated 2,600 interactions with AI Overview and AI Mode between May and July, concluding the tools failed to detect crises in 58 % (Overview) and 77 % (Mode) of 214 red‑line prompts and gave inconsistent historical answers in 43 % of cases [2]. Google countered that many of those queries were “ambiguous and contrived,” that the company could not replicate the highlighted responses, and that its own testing produced higher‑quality answers for the same prompts [1][2]. The firm also emphasized built‑in safeguards such as crisis hotlines, disclaimer notices, and parental controls that let families block the features on Android and Chrome devices [1].

Market and competitive context

Google’s AI Search tools sit alongside other generative offerings such as Meta’s AI, Anthropic’s Claude, and OpenAI’s ChatGPT, all of which face scrutiny over child safety. Common Sense Media’s broader review labeled Meta AI, Grok and several chatbot products “unacceptable risks,” while rating Gemini‑K12 and other Gemini variants as “high risk” and Claude as “moderate risk” [1]. Google’s defense highlights that, unlike its Gemini chatbot which is already limited for minors, its Search‑based AI features remain enabled by default on Android phones and Chromebooks that serve the majority of U.S. tweens and teens—a fact that amplifies the stakes of the safety debate [2].

What to watch

• Parental‑control rollout: Google’s ability to block AI Overview and AI Mode via Family Link and Chrome settings may be tested in upcoming OS updates.
• Further testing: Independent researchers may repeat or expand the 2,600‑query study, potentially prompting regulatory attention.
• Competitor responses: Rival firms may adjust their own child‑safety disclosures or limit AI features for minors in reaction to the heightened scrutiny.

The dispute underscores a broader tension: as generative AI becomes embedded in default search experiences, the line between useful educational tools and potential hazards for young users remains contested, leaving regulators and families to navigate an evolving safety landscape.