Sir Ken McCallum, the head of MI5, the UK’s intelligence service, has warned that artificial intelligence (AI) has the potential to “change the stakes” regarding existing and new bioweapons. This warning was delivered in his annual speech on threats to the nation.
Experts suggest that AI could assist various actors in developing bioweapons. For rogue, unskilled individuals or terrorist groups, AI chatbots could provide information, help plan experiments, and troubleshoot problems, likely focusing on existing pathogens and toxins.
For well-resourced organisations and nation states with skilled scientists and advanced laboratories, AI tools could help identify genetic modifications to make existing pathogens more lethal. Dr Filippa Lentzos of King’s College London stated that in this context, AI amplifies existing expertise rather than replacing it.
A more distant risk involves highly skilled groups using AI to design entirely new pathogens and toxins that are resistant to drugs or lack antidotes. Concerns have also been raised by instances such as a suspected bioterrorism plot in India involving AI chatbot queries for ricin production, and reports from AI company Anthropic of scientists attempting to use its chatbot for potentially dangerous work on viruses or toxins.
Despite these concerns, significant hurdles remain in bioweapons development. Experts highlight that information alone is insufficient, with tacit expertise, access to biological materials and equipment, experimental validation, scale-up, and delivery all presenting substantial barriers.