AI can design viruses, toxins and other bioweapons. How worried should we be? (www.nature.com)

Recent developments in artificial intelligence (AI) have raised concerns over biosecurity after Chinese scientists announced their use of AI to design conotoxins—proteins from cone snail venom that can block ion channels in the nervous system. While researchers, including the study's co-author Weiwei Xue, assert that their work aims to find therapeutic applications, a U.S. government employee warned it could pose biosecurity risks, especially since the underlying model is based on open-source research from U.S. scientists. The broader concern lies in the capabilities of AI tools that can potentially enable both amateur and sophisticated actors to design potent toxins, viruses, or bioweapons with relative ease. Experts are divided on how to manage these risks. Some advocate for stricter regulations on biological AI, while others believe the focus should be on enhancing detection methods for bioweapons. A report from the U.S. National Academies stated that while AI may help in quickly developing bioweapons, significant barriers still exist in effectively synthesizing and weaponizing these designs. With AI tools continuing to advance, the dialogue is shifting toward monitoring and mitigating the risks associated with AI in biological contexts rather than restricting innovation, highlighting the delicate balance between scientific progress and safety.