Made to order bioweapon? AI-designed toxins slip through safety checks used by companies selling genes.

The article focuses on a study conducted by researchers, including a Microsoft bioengineer, that exposed vulnerabilities in biosecurity measures designed to prevent the misuse of synthetic DNA for bioweapons. The researchers used artificial intelligence (AI) to create digital blueprints for proteins that could mimic toxic substances, such as ricin and botulinum, and found that existing screening software failed to flag many of these potentially dangerous sequences. Although improvements were made to the screening tools, concerns remain about the adequacy of biosecurity practices, particularly as AI capabilities advance. The study highlights the need for ongoing enhancements in both DNA synthesis screening and the safeguards surrounding AI protein design to mitigate risks associated with bioweapons. [Extracted from the article]