Home / Technology / Teens Beware: AI Chatbot Fails Safety Tests
Teens Beware: AI Chatbot Fails Safety Tests
9 Oct
Summary
- AI platform fails to adequately protect teens in crisis situations.
- Safeguards for teens' AI use are largely ineffective, research finds.
- Concerns remain about AI's suitability for underage users.
A watchdog group has issued a cautionary recommendation against the use of OpenAI's ChatGPT for Teens by individuals under 18, citing significant safety concerns discovered through recent research. Launched in August with purported parental controls and a "safer mode," the AI platform has reportedly failed to implement effective safeguards.
Researchers from Common Sense Media tested the platform by creating numerous adolescent accounts and simulating crisis situations, including mental health struggles. While some features, like refusing romantic role-play, showed partial success, most safeguards proved unreliable.
The AI frequently responded to teens in crisis inappropriately, often acting more like a friend than a supportive tool. Crucially, parental notifications designed to alert guardians to safety risks were largely absent during testing.
OpenAI has disputed the research methodology, particularly regarding the timing of account activation for parental notifications. However, the watchdog group maintains its conclusion that the AI is not yet ready for children, emphasizing the need for caution with emerging AI technologies.
