Google’s Gemini chatbot has been rated “high risk” for children and teenagers by non-profit Common Sense Media, which found the AI can share inappropriate sexual content, drug information, and unsafe mental health suggestions even with youth protections enabled.

The assessment adds Google to a growing list of major tech companies whose AI chatbots have shown problematic behaviour with young users, prompting industry-wide concerns about child safety in AI interactions.

Common Sense Media’s analysis tested both Gemini Under 13 and standard Gemini accounts with teen protections enabled.

Despite these safeguards, researchers found the chatbot “still expose[s] kids to some inappropriate material and fail[s] to recognise serious mental health symptoms.”

The report documented several concerning patterns:

  • Sexual content filters failed to work consistently, with some responses containing “vivid explanations of sexual content”
  • Drug-related filters were irregularly triggered, occasionally providing instructions on obtaining marijuana, ecstasy, Adderall, and LSD
  • Mental health responses were deemed potentially unsafe for vulnerable young users
  • Content complexity often exceeded age-appropriate levels for children under 13

Google’s issues mirror similar problems at competitors.

OpenAI recently announced upcoming parental controls for ChatGPT and an alert system for when young users show signs of acute distress.

Meta has modified its Meta AI to avoid discussing eating disorders, self-harm, suicide, and romantic topics with teen users.

The pattern suggests AI companies are struggling to balance open conversational capabilities with age-appropriate safety measures, particularly as chatbots become more sophisticated and widely accessible.

Common Sense Media recommends that Gemini Under 13 should only be used under strict adult supervision and advises that no user under 18 should use chatbots for mental health advice or emotional support.

The organisation urges parents to actively monitor their children’s AI usage and help interpret responses.

For Google, the non-profit suggests several improvements:

  • Better calibration of responses for specific age groups
  • Extensive testing with actual children
  • Moving beyond simple content filters to more sophisticated safety measures

Google has not yet publicly responded to the specific findings, though the company has previously stated its commitment to developing responsible AI with appropriate safeguards for younger users.

The findings raise questions about the readiness of AI chatbots for youth audiences, particularly as these tools become integrated into educational settings and daily life.

With Australian schools increasingly exploring AI tools for education, the research highlights the need for robust safety frameworks before widespread adoption.

The challenge extends beyond simple content filtering to understanding context, recognising vulnerability, and providing age-appropriate responses, complexities that current AI systems appear to struggle with consistently.

As AI chatbots become more prevalent, the gap between their capabilities and their safety measures for young users represents a significant concern for parents, educators, and policymakers.

The Common Sense Media report suggests that despite marketing claims about safety features, current protections remain insufficient for unsupervised use by children and teenagers.