TITLE:
Commentary: Unexpected Novel Chemical Weapon Agents Designed by Innocuous Drug-Development AI (Artificial Intelligence) Algorithm
AUTHORS:
Robert B. Raffa, Joseph V. Pergolizzi Jr., Thomas Miller, Daniel Motto
KEYWORDS:
Artificial Intelligence, Drug Discovery, Chemical Weapons, Machine Learning, Generative Model, Toxicity Prediction
JOURNAL NAME:
Pharmacology & Pharmacy,
Vol.13 No.7,
July
25,
2022
ABSTRACT: Recent publications reveal the disturbing information that a minor edit to an algorithm being used for designing legitimate drug candidates redirected the program in a way that resulted in the surprising design of novel chemical warfare agent candidates. Although this outcome was not the result of nefarious intent, and appropriate chemical defense authorities were notified, the potential implications of some misapplication of a drug-design algorithm for nefarious purposes are clear. This Commentary summarizes how otherwise benign Artificial Intelligence (AI) algorithms used for drug discovery can be easily reversed to design novel chemical warfare agents for which no effective antidote will be available, or perhaps even envisioned.