MI5 chief warns AI could help terrorists develop deadly new bioweapons if guardrails fail

Sir Ken McCallum, director general of MI5, has warned that AI has the potential to "change the stakes" on bioweapons, giving both lone terrorists and hostile nation states a dangerous new edge. The warning came during his yearly address on national security threats, where he flagged AI chatbots as tools that could help unskilled actors rapidly acquire dangerous capabilities using existing pathogens. Dr Filippa Lentzos of King's College London warns the more immediate danger is AI amplifying the expertise of well-resourced state actors. Indian police have already uncovered a real-world case: a suspected bioterrorism plot involving a doctor who consulted AI chatbots.

McCallum warned AI could help pinpoint genetic modifications that would render existing pathogens deadlier.

The most distant — but gravest — risk is that sophisticated groups could exploit AI to engineer entirely new pathogens for which no antidotes exist.

The MI5 chief's intervention came as part of his annual national-threat address, covering risks from hostile states, terrorists and criminal networks.