AI Drug Discovery Systems Might Be Repurposed to Make Chemical Weapons, Researchers Warn

2020 Collaborations Pharmaceuticals, a company that specializes in looking for new drug candidates for rare and communicable diseases, received an unusual request. The private Raleigh, N.C., firm was asked to make a presentation at an international conference on chemical and biological weapons. Artificial intelligence software can be used to develop drugs for treating diseases such as Pitt-Hopkins syndrome or Chagas disease.

Sean Ekins, the chief executive of Collaborations, began to work with a senior scientist at the company. Collaborations came up with an idea that could be used to create a compendium of toxic compounds that were similar to the notorious nerve VX.

The team ran MegaSyn overnight and came up with 40,000 substances, including not only VX but other known chemical weapons, as well as many completely new potentially toxic substances. It took a bit of programming, open-source data, a 2015 Mac computer and less than six hours of machine time.

Spiez CONVERGENCE, a conference in Switzerland that is held every two years to assess new trends in biological and chemical research that might pose threats to national security was presented by collaboration. The company's research was published in the journal Nature Machine Intelligence and they gave a report on the findings to the White House Office of Science and Technology Policy.

The resemblance to the company's day-to-day routine work was startling. The researchers used MegaSyn to create the same molecule with the same target. Alzheimer's can be treated with these drugs, called acetylcholinesterase inhibitors. The software was asked to create substances similar to VX without putting the exact structure of the molecule.

Alex MacKerell, director of MegaSyn, says artificial neural networks are telling us which roads to take to a specific destination. The score is a molecule based on certain criteria, such as how well it works. A higher score indicates that the substance might have a better effect.

According to the company's scoring method, many of the novel molecule MegaSyn were predicted to be more toxic than the one used in VX, which made both Ekins and Urbina uncomfortable. They wondered if they had crossed an ethical boundary by running the program and not doing anything more to narrow down the results.

The director of the Center for Ethics at Emory University, who was not involved in the research, says that their ethical intuition was right. Artificial intelligence is one of the issues that Wolpe frequently writes about. The authors had to not push it any further once they felt they could demonstrate that this was a threat.

Some experts say that the research didn't answer important questions about whether using artificial intelligence to find toxins could lead to the development of a biological weapon.

The development of actual weapons in past weapons programs has shown that what seems possible theoretically may not be possible in practice.

Elana Fertig, an associate director of quantitative sciences at the Sidney Kimmel Comprehensive Cancer Center at the John'sHopkins University, says that the ease with which an artificial intelligence can rapidly generate a vast quantity of potentially hazardous substances could still speed up the process of creating lethal bioweapons

The authors of the paper propose a number of ways to make it harder for people to misuse these technologies, including wait lists that would require users to undergo a prescreening process to verify their credentials before they could access models.

Drug discovery artificial intelligences can be presented to the public through an application programming interface. A user would have to specifically request the data. In an e-mail to Scientific American, Ekins wrote that an application programming interface could only be structured to generate molecule that would minimize potential toxicity and demand the users apply the tools/models in a specific way. Ben Ouagrham-Gormley believes that the technology could be used to foster bio weapon development.

They don't want to portray these things as being bad because they actually do have a lot of value. I think it is important to consider that note of caution.