Annie Jacobsen
We are standing on a razor’s edge. Artificial intelligence is reshaping the biological sciences with terrifying speed. It cuts two ways. One path leads to disaster: bad actors using prompts to generate biological weapon recipes. The other path offers salvation. AI can track disease outbreaks in real time, pushing critical health data to millions before the first hospital bed is filled. This is not a theoretical debate. It is a race. A race between the proliferation of “dark biology” and a new era of collective global health protection.
Detecting outbreaks before they spiral
Every epidemic starts small. A handful of cases. Maybe a few weird symptoms in a remote village. Public health wins only if we learn about the outbreak early. Before hospitals overflow. Before containment becomes impossible. As transmission rates explode, the options shrink. Fast.
Here is where AI matters most. It compresses the timeline between the first sign of a disease and the moment we know it’s there. Speed is the difference between a contained cluster and a global pandemic.
The World Health Organization (WHO) already understands this. At its Hub for Pandemic and Epidemic Intelligence in Berlin, they are running computational systems designed to monitor infectious diseases worldwide. The centerpiece is a system called Eios (“EE-oss”).
Epidemic Intelligence from Open Sources.
For decades, we relied on official data. Case counts from hospitals. Lab reports. Official databases. The problem? This data takes days, sometimes weeks, to filter through bureaucracy. By the time it’s official, the virus has often spread.
Eios changes the game. It uses AI to scan millions of unofficial sources. Social media chatter. Local news reports. Citizen posts. It operates in multiple languages, scraping the web for signals of disease. It filters the noise. It gives us a near-instant picture of where outbreaks are starting.
The lowering barrier to dangerous science
AI is a force multiplier. It is changing biology every single day. The real question isn’t if AI can cause harm. It can. The question is whether defense can move faster than offense.
The offensive capability is already here. AI systems can navigate massive libraries of biological knowledge. They can explain complex lab techniques. They can troubleshoot experiments. They can suggest new designs. Individually, these tools seem benign. In combination? They lower the barrier to entry for a field that was once reserved for elite specialists.
The fear? That AI will help humans create entirely new forms of life. Mirror-image organisms. Pathogens with capabilities never seen in nature. This stuff used to be science fiction. Now, it’s experimental biology.
Why biological threats are unique
Biology behaves differently than software or hardware. A software bug crashes your laptop. You can pull the plug. A biological error? It can replicate. It can mutate. It can evolve. It can cross borders.
Once a living organism is released, you can’t just hit “recall.” It doesn’t work that way. This fundamental difference makes biological weapons more dangerous than arrows or even intercontinental ballistic missiles. They are alive. They have agency, of a sort.
The barrier to entry for creating these threats is incredibly low. Nuclear weapons require enriched fissile material. Complex delivery systems. State-level resources. Biological weapons? They can be engineered with widely available lab tools. Enhanced by increasingly powerful, accessible AI.
Closing the gap between design and defense
The technology is advancing faster than our laws. Our treaties. Our safeguards. That widening gap is where the existential danger lives. We don’t need panic. We need pressure. We need action.
We must deploy AI as a defensive shield. We need systems that accelerate detection. Strengthen public health infrastructure. Build biological defenses at the same pace that AI is accelerating biological design.
Because if biology outruns our ability to contain it… there is no turning back.
