Microsoft says AI can create “zero day” threats in biology

Microsoft researchers have used AI to identify a vulnerability in biosecurity systems meant to prevent DNA misuse, potentially allowing the creation of harmful proteins. The team demonstrated how generative AI algorithms could bypass screening software used to detect dangerous genetic sequences. While the study was purely digital and no toxic proteins were produced, it highlighted the need for enhanced screening procedures in DNA synthesis. The findings underscore the importance of integrating biosecurity measures into AI systems to prevent misuse. The research has prompted calls for improved security measures in DNA synthesis to counter potential bioterror threats.