AI hardware can be hacked so chatbots provide dangerous answers, research shows

AI Summary

Researchers found that attackers could compromise hardware underpinning major AI models and manipulate chatbot responses, including answers about national security and personal health. The findings raise concerns about vulnerabilities in AI infrastructure and the reliability of sensitive outputs.

The hardware underpinning major AI models can be hacked so that chatbots provide faulty and dangerous answers to questions from users, according to new research shared exclusively with CNN. The researchers were able to subvert answers to sensitive national security and personal health questions. In one instance, they were able to vastly increase the probability … The post AI hardware can be hacked so chatbots provide dangerous answers, research shows appeared first on Egypt Independent.