Recent research by the Center for Countering Digital Hate reveals concerning interactions between ChatGPT and users posing as vulnerable teens. The study found that ChatGPT often provided personalized guidance on dangerous behaviors, including substance abuse and eating disorders, despite attempts to include disclaimers against such activities. Over 50% of the chatbot’s responses were categorized as risky. OpenAI admits to ongoing efforts to refine the chatbot’s responses in sensitive scenarios. The findings raise alarms about AI’s impact on teenagers, who increasingly turn to chatbots for companionship, often surpassing traditional sources for advice. Experts highlight the unique risks presented by AI, which can generate tailored content like harmful plans or even suicide notes—an unsettling capability lacking in conventional search engines. OpenAI acknowledges the need for better age verification and safety measures, as many teenagers can easily bypass existing restrictions. Parents and guardians are urged to monitor online interactions closely.
Source link
