Most AI chatbots easily tricked into giving dangerous responses, study finds

By admin

Researchers say threat from ‘jailbroken’ chatbots trained to churn out illegal information is ‘tangible and concerning’