A new study reveals AI can bypass biosecurity measures used by DNA manufacturers, allowing potential misuse of dangerous genetic material. Researchers demonstrated AI protein-design tools could 'paraphrase' toxic gene codes, evading existing safety screens. While a fix was implemented, it's not foolproof. This highlights growing concerns about AI's role in creating biothreats, prompting scientists to restrict data sharing and develop new risk management models for sensitive research.

Prepared by Olivia Bennett and reviewed by editorial team.