Tablets believed to be laced with fentanyl are displayed at the Drug Enforcement Administration ... [+]
A team of global experts in AI and synthetic chemistry have just published their research into AI’s capacity to identify hundreds of ways to scale manufacturing of fentanyl that evade current DEA controls. The implications are striking.
The study’s implications suggest that if only a fraction of the information provided progresses to its logical conclusion, the already out-of-control fentanyl crisis, will be supercharged and could present an imminent threat—a real-life Breaking Bad.
What makes AI-assisted illegal drug synthesis so dangerous, according to the Chem-Catalysis piece is that, using decades old AI technology, it is possible to identify scores of pathways for production of fentanyl using readily available precursor materials. These AI-algorithmically scripted formulas or “recipes” are extraordinarily inexpensive to “cook.”
The lead author, Dr. Bartosz A. Grzybowski, Distinguished Professor, Department of Chemistry, UNIST Director, Center for Algorithmic and Robotized Synthesis (CARS), Institute for Basic Science based in Ulsan, Republic of Korea, stated in a phone interview: “We did this piece as a first step in alerting governmental bodies and the global public as to the current power of AI in drug manufacturing. AI’s power is by no means limited to drug manufacturing and we should now begin the process of understanding the positive and negative possibilities of AI technologies.”
Dr. Grzybowski and his coauthors are chemists and AI specialists who have worked on the development of chemical AI for over a decade. According to Grybowski, setting up the computer protocol took only two days and produced the machine-learning solutions in 20-minutes. “The longest part of the effort was writing the journal article, which took two-months.”
Another warning in a companion piece by Dr. Rick Bright, stated that “The U.S. should not simply prepare for just another drug wave.” Bright, one of the Chem-Catalysis piece reviewers is also chief executive of Bright Global Health and former deputy assistant secretary for preparedness and response and director of BARDA at the US Department of Health and Human Services.
One need not be a world leader in AI-assisted pharmacology to start illicit drug production. The paper outlines the key precursor chemicals such as N-Phenethyl-4-piperidone (NPP), 4-anilino-N-phenethylpiperidine (ANPP), norfentanyl, N-Phenyl-4-piperidinamine (4-AP), and tert-Butyl 4-(phenylamino) piperidine-1-carboxylate (1-boc-4-AP). It is possible to obtain all the precursor chemicals legally. Not mentioned in the article, is the fear that the EPA will be unable to even monitor the risks created by the countless number of lethal byproducts, the carcinogens, the waste stream that might be generated.
Given that there are multiple drug recipes, there will not be just a few supply chain routes; there will be countless supply chain highways. Even though the DEA seized over 77 million fentanyl pills and 6 tons of fentanyl powder in 2023, the war on drugs is far from over. In 2023 alone, there were 111,355 reported opioid-related deaths in the U.S. Add to that the 640,000 opioid-related deaths in the world, and one gets an idea of the problem’s scope given that only a fraction of opioid-related deaths are reported.
The study’s implications suggest that if only a fraction of the information provided progresses to its logical conclusion, the already out-of-control fentanyl crisis, could be supercharged. Preparing for the next pandemic is important, but, if these experts are correct, the next pandemic might not be caused by a pathogen; rather, it may very well be facilitated, if not caused, by AI through employment of AI-driven chemical synthesis for the manufacturing of fentanyl or other opioid derivatives.
As a simple primer, Fentanyl is a synthetic opioid 50 times more potent than heroin and 100 times more potent than morphine. Currently, China-sourced fentanyl seized by U.S. law enforcement is over 90% pure. AI-assistance has the potential to refine illicit fentanyl production and increase its purity making it even more deadly.
AI technologies have the potential to both ameliorate and exacerbate the situation. Pharmaceutical companies are already using AI to maximize revenue and optimize patient usage by finding new ways to develop, synthesize, manufacture and distribute drugs with unprecedented efficiency and flexibility.
Imagine that same power in the hands of illicit drug manufacturers. Prior to the current fentanyl surge, other drugs have hit the U.S. in smaller waves. In the late 1800s and early 1900s, it was heroin. Next came use of the stimulant cocaine in the 1970s. The advent of a smokable form known as crack cocaine expanded its use in the 1980s. The early 2000’s saw the rise of methamphetamine, spawned by remedial chemists and the ease of domestic manufacturing created from readily available precursor ingredients.
The 21st century ushered in another opiate epidemic, this time driven by the over-prescription of synthetic opioids like fentanyl and oxycodone. Heroin use also increased, as addicts turned to the cheaper and more potent drug. Heroin use nearly tripled between 2007 (161,000) and 2014 (435,000).
Today, illicit fentanyl taints all statistics on individual recreational drug usage because current supplies of heroin, cocaine and methamphetamine contain varying doses of the synthetic opioid. The practice of adding cheap fentanyl to the more expensive drugs increases both potency and profit margin. According to the DEA, a staggering 6 in 10 fake prescription pills contain potentially lethal doses of fentanyl.
One kilogram of fentanyl has the potential to kill 500,000 people. A lethal amount of fentanyl is barely visible next to the penny below:
A penny next to 2 milligrams of fentanyl, a lethal dose for most people, according to the DEA
According to an April 8, 2024, article in The Economist, drug networks are well ahead of authorities in the fight over Europe’s ports. The ports are considered relatively safe drug hubs as agents can inspect only 1.5% of containers. In the U.S., the number is about 2% of the 750 million containers arriving each year.
The reality is that the current supply chain for opioids and stimulants shows no signs of disruption. With AI-facilitated synthetic opioid production looming, the problem will get worse. Grzybowski and associates speculate that there will be less of a need to import fentanyl and fentanyl-laced drugs if “cooks” start using AI algorithms to establish their cottage industries on U.S. soil.
An increase in supply will lower costs, meaning that individuals with substance use disorder will have an easier time accessing the drugs. Unknowingly, those who assume they are consuming a “safer” drug like heroin, cocaine or methamphetamine will very likely be ingesting fentanyl as well. According to the CDC, U.S. drug overdose deaths have soared from 8.2 per 100,000 in 2002 to 32.6 per 100,000 in 2022.
Overdose patients fortunate enough to receive an opioid antagonist such as naltrexone in the field go to the local emergency room for continued treatment. Once a patient is deemed “stable,” they leave the E.R. with instructions for outpatient follow-up care. Unmonitored, such a patient is at risk of further injury because there are drawbacks to current opioid antagonists. Their administration can plunge a patient into opioid withdrawal sickness. Without support, the person who uses drugs will often seek more, or more potent, drugs for relief. In a large study, 8.3% of previously discharged “stable” patients died within 3 days of the original overdose.
Another problem with opioid antagonists is their duration of action. Onset is rapid (2-3 minutes), but the effects last only 20 minutes, wearing off much faster than the synthetic opioids. For example, depending on drug purity and dosage, illicit fentanyl’s effects linger for hours. In overdose situations, treatment may require repeat dosing or continuous infusion of the opioid antagonist to counteract synthetic opioid effects.
Based on the Grzybowski team’s article, it is realistic to assume that:
· The current flow of illegal drugs will continue.
· The flow of drugs to market will grow due to AI’s ability to facilitate large-scale production and distribution.
· Fentanyl and fentanyl-laced drugs will predominate.
· Opioid drug overdoses will increase.
· A surge in overdose cases could easily overwhelm the already taxed system of emergency services.
Emergency rooms nationwide already operate above maximal capacity leading to overcrowding, ambulance diversions and challenges in managing patient flow. AI-facilitated illicit drug production will make the situation worse. When that happens, what will be the effect on other patients in need of emergency care? Where will the accident victim, the heart attack patient and the acute stroke patient go for care when the local emergency room is overwhelmed with drug overdoses?
Covid-19 has already taught that the current U.S. healthcare infrastructure cannot accommodate surges. To avoid, or at least better handle, a surge will require not only life-saving rapid onset medications, but also medications with a sufficient duration of action to counter the effects of synthetic opioids. Only with such therapies will it be possible to avoid plunging a patient into opioid-withdrawal sickness and/or recurrent overdose danger.
AI influence in healthcare has been a mixed bag. At its best, it will streamline clinical trials, accelerate drug development (especially through better molecular folding processes), boost manufacturing by improving raw material acquisition and processing, simplify distribution chains and reduce bottle-necking and much more.
However, the mishandling of AI technology can be equally powerful. Advantageous manufacturing innovations can be highjacked to produce illegal drugs like fentanyl, at a scale that will dwarf existing levels producing unimaginable devastation. Dr. Bright closed by saying, “We must prepare for an illicit drug surge as great a threat as that posed by a relentless rise in sea levels that breach all barriers.”
