AI, Bioweapons, and the Future of Warfare

Danny Crichton:

AI can bring out the nightmares in all of us. Displacing workers, seizing control of the planet, annoying us as virtual pets. But let's add one more potential nightmare to the list Bioweapons. Can I make biological and chemical weapons even more deadly? Let's take a look.

This debate got started last year with a paper by biologist Fabio Urbina, who explored what would happen if artificial intelligence tools for drug discovery were repurposed to find the most lethal toxins to the human body. Unsurprisingly, results were immediate and grim. Quoting from his team's report, In less than 6 hours after starting on our in-house server, our model generated 40,000 molecules that score within our desired threshold.

In the process, the A.I. designed not only the nerve agent, the VX but also many other known chemical warfare agents that we identify through visual confirmation with structures and public chemistry databases. Many new molecules were also designed that looked equally plausible. These new molecules were predicted to be more toxic than publicly known chemical warfare agents. It's a scary and plausible paper.

After all, many countries throughout history have developed bio and chemical weapons. The United States just recently completed the final destruction of its last chemical weapons supplies. While the implications of the article are indeed terrifying, they are hardly surprising. In fact, bioweapons and to a much lesser extent chemical warfare have been considered the top U.S. national security threat for more than two decades now.

There have been intense concerns that synthetic biology, and particularly the development of DNA editing tools like CRISPR, could lead to an exponential increase in bioweapon threats. It makes sense. After all, the precision, coupled with greater democratization, portended a rapid rise in the capability of bad actors to invent horrific new pathogens to inflict on our world. Yet if we take a step back, we realize that such potential is more imaginative than substantive.

In their Nature article, Urbina and his team note that they discovered many compounds more toxic than VX nerve gas. But how toxic is VX in the first place? Urbina writes that a few salt sized grains of nerve gas is sufficient to kill a person. It's essentially as deadly as a substance can possibly be going for a few grains of toxin to reach lethality to a single grain of toxin is hardly a major qualitative advance.

That's indeed a recurring pattern in this field. While there are widespread fears of mad scientists inventing deadly contagions and hidden wet labs in the caves of Waziristan, the reality is that the world is already familiar with incredibly viral and deadly pathogens. Ebola kills roughly half of anyone infected with a relatively high virality rate. As one former presidential adviser on bioweapons explained to me years ago, Mother Nature is quite efficient at producing terrifying bioweapons on her own.

No mad scientist required. Our public health response to a naturally occurring pandemic and one that is man made will be exactly the same. All that critique aside, there was one element of Urbina’s tale where it got a bit more concerned. Writing on the potential for using artificial intelligence to find new manufacturing pathways for chemicals. Urbina wrote that, quote, We did not assess the virtual molecules for synthesizeability or explore how to make them with retro synthesis software for both of these processes.

Commercial and open source software is readily available that can be easily plugged in to the de novo design process of new molecules with current breakthroughs and research into autonomous synthesis. A complete design make test cycle applicable to making not only drugs but toxins is within reach. Our proof of concept does highlight how a non-human autonomous creator of a deadly chemical weapon is entirely feasible.

One of the only checks on chemical weapons production is that large scale manufacturing is typically recognizable from satellite imaging as well as from the purchase of manufacturing equipment and chemical precursors. Since we know how these weapons are made, we can search for the right clues to indicate their manufacture. Unfortunately, this is less relevant to bioweapons since they are self propagating and thus don't give off the same scale related signals.

With new AI tools, it's not just that alternative pathways can make it simpler to produce these weapons, but that a covert program could hide its tracks by simultaneously using different pathways to obfuscate its true intentions. If there are 100 unique ways to reduce VX nerve gas, it gets progressively harder to track. AI won't make bio and chemical weapons more lethal, but it could make them easier to manufacture.

So let's add one more doomsday nightmare to the AI bucket list for now.