For most of us, artificial intelligence conjures images of chatbots helping with emails or recommending what to watch next. But a growing number of biosecurity experts are warning that the same technology could, in the wrong hands, lower the barrier to something far more serious: a deliberately engineered pandemic.
It sounds like the plot of a thriller. Unfortunately, it’s a scenario now being taken seriously by scientists, security researchers and government advisors around the world.
What’s actually being warned about
Creating a virus capable of causing mass harm has traditionally required years of specialist training, laboratory access and technical know-how that only a small number of people in the world possess. That’s been one of our main defences against bioterrorism: even highly motivated groups have generally lacked the expertise to pull it off.
The concern now is that AI systems are eroding that barrier. Researchers at institutions including MIT have tested whether freely available AI chatbots could guide someone with no scientific background through the early stages of identifying dangerous pathogens and working out how their genetic material might be obtained. In more than one study, the answer has been uncomfortable: with the right prompts, some AI tools have offered up detailed and specific suggestions.
It’s worth being clear about what this does and doesn’t mean. Having a list of ideas from a chatbot is a long way from actually producing and releasing a pathogen capable of starting a pandemic. You’d still need to synthesise genetic material, get it assembled by a lab, and then find a way to produce and spread it effectively — each of those steps has its own technical hurdles and, in many cases, safety screening designed to catch exactly this kind of misuse. Independent assessments, including a well-known study by the RAND Corporation, initially found that AI assistance didn’t give would-be attackers a meaningful edge over a good internet search. However, more recent reviews suggest that gap is narrowing as AI models become more capable, and that the trend is worth watching closely rather than dismissing.
Why this matters for ordinary preparedness
I want to be careful here not to overstate the risk. A successful large-scale bioterrorism attack remains, thankfully, extremely rare — the historical record shows only a tiny handful of attempts using biological agents out of hundreds of thousands of terrorist incidents worldwide. Most experts who study this closely describe the risk as growing rather than imminent, and stress that better safeguards around AI models and DNA synthesis companies can meaningfully reduce it.
But this is exactly the kind of low-probability, high-consequence scenario that sits at the heart of sensible preparedness thinking. We don’t prepare for pandemics because we expect a deliberate attack tomorrow. We prepare because the practical response to any severe disease outbreak — whether it emerges naturally, escapes from a laboratory, or is caused deliberately — looks remarkably similar in the first days and weeks: a period of uncertainty, possible movement restrictions, disruption to supply chains, and pressure on healthcare services.
What this changes, practically speaking
For most households, the honest answer is: not very much changes day to day. The fundamentals of resilience remain the same regardless of how a future pandemic might start:
- A reasonable store of food, water and essential medication that would see your household through several weeks without needing to rely on shops or deliveries
- A basic understanding of how infectious diseases spread, and simple hygiene and isolation measures you could put in place quickly at home
- A plan for how you’d manage if schools closed, workplaces went remote, or local movement was restricted for a period
- Staying reasonably informed through official channels during any emerging outbreak, rather than relying solely on social media speculation
What this news does is add another card to the deck of reasons why pandemic preparedness deserves a place in ordinary household planning, alongside more familiar risks like severe weather or power disruption. It’s not about living in fear of a specific threat. It’s about recognising that the systems we depend on — food supply chains, healthcare capacity, information networks — can come under strain from more than one direction, and building in a sensible margin of resilience regardless of the cause.
Whether the next serious outbreak originates in a wet market, a research laboratory, or is engineered deliberately, the households that come through it most comfortably will be the ones who already had the basics sorted before it started. Get The Book – Preparing for Lockdown