AI Hardware Hack Could Make Chatbots Give Dangerous Answers, Study Finds
Part of composite article AI Hardware Hack Lets Attackers Rig Chatbots to Give Dangerous Answers, Study Finds View full article →
The physical hardware that powers major artificial intelligence models can be hacked to make chatbots give faulty and dangerous answers to users, according to new research shared exclusively with CNN. Researchers were able to manipulate answers to sensitive national security and personal health questions. In one case, they greatly increased the probability of a harmful response.