Scientists develop 'toxic AI' that comes up with 'harmful' answers

Researchers at MIT are training AI chatbots to generate harmful and hateful responses as a method of tackling toxic content by using the prompts to filter out dangerous responses

Related Keywords

United States ,France ,Mulhouse ,Alsace ,Pulkit Agrawal ,Improbable Ai Lab ,Artificial Intelligence ,Getty Images ,Curiosity Driven Red Teaming ,Massachusetts Institute Of Technology Mit ,