Child safety concerns are significant in Google Search AI features: report
text_fields
Google Search's artificial intelligence features pose an "unacceptable risk" to children, according to a report by the Youth AI Safety Institute at Common Sense Media, which found gaps in child safety, mental health responses and content moderation.
The study tested more than 2,500 queries on Google Search's AI Overviews and AI Mode using accounts configured for minors.
Researchers said the AI failed to recognise suicidal ideation, normalised an eating disorder symptom, recommended an outdated helpline and provided guidance related to deepfakes.
The report also said the AI features are enabled by default and cannot currently be disabled by parents or schools, though Google Search itself can be turned off for child accounts.
Google rejected the findings, saying the report relied on a limited set of queries that did not reflect typical user behaviour. The company said its AI search features include additional safety measures and that follow-up questions help the AI better understand user intent.
The report did not assess Google's standalone Gemini chatbot or its third-party integrations.