ChatGPT sometimes gives dangerous advice to teens: Study
A recent study found that ChatGPT, the popular AI chatbot, sometimes gave teens dangerous advice on tough topics like drinking, eating disorders, and suicide.
Researchers at the Center for Countering Digital Hate (CCDH) sent 1,200 prompts based on real mental health struggles.
Shockingly, over half of the responses included unsafe tips—like how to hide an eating disorder or get drunk.
ChatGPT's safety features easily tricked
The researchers managed to trick ChatGPT's safety features just by rewording their questions as if they were for a friend or a school project.
In some cases, ChatGPT even wrote detailed plans or personal letters about suicide.
CCDH's CEO Imran Ahmed called the chatbot's protections "barely there," highlighting how easily young users could get harmful info.
Experts call for urgent need for better age checks
With more than 70% of US teens using AI chatbots regularly, experts say it's urgent to make these tools safer.
They're pushing for better age checks and stronger safeguards so that young people aren't put at risk when they turn to AI for help.