By Taylor Hinds
Addiction is not a moral failing, nor a lack of willpower, nor or a character flaw. It is a chronic, relapsing neurological disorder in which the brain’s reward circuitry, which has been shaped by millions of years of evolution to seek food, shelter, and connection, is hijacked by substances and behaviors that exploit those same ancient pathways. Understanding this fundamental truth is essential to understanding why, until very recently, we have been so bad at treating it.
Across the arc of human history, addiction treatment has largely been a story of limitation, stigma, and modest results. But something unprecedented is now happening in laboratories and clinics around the world. A class of medicines, Health Optimizer Therapies (HOTs), led by GLP-1 receptor agonists like semaglutide, appear to quiet the very reward circuitry that addiction exploits by recalibrating the brain’s relationship with craving itself.
The implications, if borne out by the clinical trials now underway, could be extraordinary. Not just for the estimated 50 million Americans currently struggling with substance use disorders, but for the broader vision of what human health and potential can look like in the 21st century.
A History of Damage Control
For most of recorded human history, addiction treatment was indistinguishable from moral condemnation. The drunk, the drug addict and the smoker were all were understood primarily through the lens of personal weakness or spiritual failure. The dominant treatment was abstinence, enforced through willpower, religious conversion, or incarceration. Relapse was proof of moral inadequacy instead of the expected behavior of a chronic neurological condition.
The 20th century brought genuine scientific progress. Researchers mapped the brain’s mesolimbic dopamine system, the reward pathway, and began to understand how addictive substances hijack it. This led to the first generation of pharmacological interventions, each targeting specific aspects of the addiction cycle.
Methadone maintenance therapy, introduced in the 1960s, offered heroin users a longer-acting opioid substitute that could be taken orally, stabilizing their lives and reducing crime. It worked, but required daily supervised dosing at specialized clinics, creating enormous barriers to access and carrying its own risks of dependency.
Naltrexone (approved 1984 for opioids, 1994 for alcohol) and naloxone (approved 1971) arrived as antagonist therapies, blocking opioid receptors so that drugs produced no euphoric effect. Effective in principle, but reliant on a patient continuing to take a medication that prevents them from getting high. This created a compliance problem that proved nearly impossible to solve at scale.
Buprenorphine (a partial opioid agonist approved 2002) improved on methadone by allowing home administration. But it still operates on the same substitution logic to manage the disorder by partially satisfying the craving with a safer compound.
For alcohol, disulfiram (Antabuse, approved 1951) worked by making alcohol consumption physically unpleasant, inducing nausea and flushing. It, of course, was only effective when patients actually took it, which they frequently did not. Acamprosate (approved in 2004) and naltrexone offered more modest benefits, reducing but rarely eliminating relapse. Naltrexone blocks the opioid receptors that mediate alcohol's euphoric and reinforcing effects which dull the reward signal so that drinking becomes less compelling. Acamprosate targets the glutamate system by restoring balance and reducing the restless, anxious craving that dominates early recovery. In clinical trials, both roughly double the odds of sustained abstinence compared to placebo but placebo abstinence rates in alcohol trials are often below 20%. Most patients on these medications still relapse.
Nicotine replacement therapies (patches, gums, and lozenges) and varenicline (Chantix, approved 2006) roughly doubled nicotine cessation rates compared to placebo, though placebo quit rates are often below 10%. Varenicline is a medication that acts as a partial agonist at nicotine receptors, partially satisfying the craving while blocking the full reward from smoking. It is the most effective single agent, but still only achieves 12-month abstinence in 20-30% of users. Patches, gums, and lozenges all contain nicotine and are intended to help people quit smoking, not quit nicotine, and dependency often persists long past the recommended cessation period. Vapes and pouches occupy a similar position to NRTs by offering a cleaner delivery mechanisms for the same addictive molecule. This is useful perhaps as harm reduction, but not cessation in any meaningful sense, and none of these approaches address what makes nicotine compelling in the first place.
The pattern across all of these treatments is consistent. They are helpful at the margins, hard to adhere to, narrow in scope (each addresses one substance), and leave the majority of patients relapsing. As the National Institute on Drug Abuse (NIDA) has long acknowledged, untreated opioid use disorder carries a relapse rate of 80-90% within a year. Even with best-available medication-assisted treatment, that number drops only to 40-50%.
Existing addiction pharmacotherapies work by targeting the same specific receptor systems that the drugs of abuse exploit. They treat the downstream manifestation of addiction rather than the upstream neural machinery that generates craving in the first place. Furthermore, for the vast majority of addictive substances like cocaine, methamphetamine, cannabis, gambling, and compulsive digital behaviors, there are no FDA-approved medications whatsoever.
The Architecture of Craving
All addictive substances and behaviors share a common mechanism. They flood the brain's reward system with dopamine, the chemical signal that says "that was important, do it again." Over time, the brain adapts. It becomes less sensitive to dopamine, so more of the substance is needed to feel the same effect. And the places, people, and rituals associated with drug use become powerful triggers in their own right, capable of generating intense craving even in the complete absence of the substance.
This is why willpower often fails. A craving is a neurological event, driven by circuitry that operates largely below the level of conscious control.
The Unexpected Architecture of GLP-1 Receptors in the Brain
GLP-1 is a hormone naturally produced in the gut after eating. Its best-known job is to signal to the brain that you are full and to trigger insulin release, which is why drugs like semaglutide that mimic it are so effective at reducing appetite and calorie intake.
But GLP-1 doesn't only work in the digestive system. The brain also produces it, and GLP-1 receptors are found throughout the brain's reward circuitry, the same network that addiction hijacks.
When GLP-1 agonists (semaglutide, tirzepatide etc.) activate these receptors in the brain's reward system, they appear to turn down the volume on the signal that makes rewarding things feel compelling. Food, alcohol, drugs, and gambling all trigger a dopamine surge. GLP-1 agonists seem to dampen that surge, not by blocking pleasure entirely, but by quietly reducing how loudly the brain responds to it. The craving becomes less insistent and the pull becomes easier to resist.
One of the most clinically significant findings is that GLP-1 agonists appear to reduce what researchers call cue-reactivity, the way that sights, smells, and situations associated with drug use trigger powerful cravings even when the substance isn't present. Brain imaging studies showed that people given a GLP-1 drug responded less intensely to alcohol-related cues than those on placebo.
This is fundamentally different logic than every intervention that has come before. It is not replacing one substance with another. It is not preventing the drug from working at all. It is modulation of the underlying system that decides what is worth wanting in the first place. This is why users report that semaglutide reduces not just food cravings but alcohol consumption, tobacco use, opioid craving, gambling impulses, and even compulsive social media scrolling, often simultaneously, without specifically targeting any of those behaviors.
The Evidence: What We Know So Far
The most important study to date was published in The BMJ in March 2026. Researchers at Washington University School of Medicine analyzed health records from more than 600,000 US veterans with type 2 diabetes, comparing those taking GLP-1 agonists against those on other diabetes drugs. The findings were striking across every substance category tested. GLP-1 agonists were associated with 50% fewer substance-related deaths, 39% fewer drug overdoses, and 26% fewer drug-related hospitalizations. They also reduced the risk of new substance use disorders across alcohol, opioids, cocaine, cannabis, and nicotine. This is the first study to test the effect across all major addictive substances simultaneously, and the consistency of the findings across categories is what makes it so significant.
Alcohol
Randomized controlled trials have shown semaglutide reduced both craving and alcohol consumption in people with alcohol use disorder, and dulaglutide reduced drinking in a separate trial. A large Swedish registry study following over 227,000 people with alcohol use disorder over nearly nine years found that those taking semaglutide had a 36% lower risk of alcohol-related hospitalization compared to periods when they were not on the drug. This is a larger reduction than any of the medications currently approved specifically for alcohol use disorder.
Opioids
In animal models, GLP-1 agonists have been shown to reduce self-administration of heroin, fentanyl, and oxycodone, and to reduce reinstatement of drug-seeking behavior (relapse). In the BMJ study of veterans, GLP-1 users with existing opioid use disorder showed substantially fewer overdoses and drug-related deaths. Dedicated human trials are underway but have not yet reported results.
Cannabis
There are currently no approved medications for cannabis use disorder anywhere in the world. An earlier observational study found semaglutide associated with significantly reduced incidence and recurrence of cannabis use disorder. The BMJ study confirmed this signal at much larger scale.
Nicotine
The animal evidence is consistent and compelling. GLP-1 drugs reduce nicotine self-administration in rodent models, and appear to make nicotine actively aversive. In the BMJ study, GLP-1 users had fewer healthcare visits for nicotine dependence. Human trials specifically targeting smoking cessation are underway.
Cocaine and stimulants
This is the area with the greatest unmet need and the thinnest evidence. There are no approved treatments for cocaine or methamphetamine addiction. The BMJ study found a reduced risk of new cocaine use disorder among GLP-1 users, which is a meaningful signal. As of early 2026, four clinical trials of GLP-1 drugs for cocaine use disorder are registered, and one for methamphetamine.
It is important to note that most of the large studies so far are observational, not randomized controlled trials. The BMJ study used sophisticated methods to approximate the rigor of a randomized trial, but confounding cannot be fully ruled out. In general, however, the signal is now consistent enough across animal studies, observational data, randomized trials, and independent research groups around the world that the question has shifted from whether GLP-1 drugs have meaningful effects on addiction to how large those effects are, for which substances, at what doses, and for which patients.
GLP1s for Addicts
One major problem with this application of GLP-1s is that they are not approved to treat addiction, only for type 2 diabetes and obesity. Because no addiction indication exists, insurance companies will not cover them when prescribed for substance use disorders. Phase 3 trials for alcohol use disorder are underway, and a formal FDA addiction indication is the essential next step.
Even if that regulatory hurdle were cleared tomorrow, another challenge is that the patients who could most benefit are often the worst candidates under current approval criteria. GLP-1 drugs are approved for people with high BMI but many people with serious addiction are dangerously underweight. On admission for detoxification, up to 70% of people who use drugs have body weight below the normal range which is unsurprising given that opiates, cocaine and methamphetamine all suppress appetite and gut function. Users are often malnourished and physically depleted long before the question of treatment arises. Prescribing a drug whose primary mechanism is further appetite suppression to this population raises real safety concerns.
A key clinical question is whether the neurological benefits, dampening craving and recalibrating the reward system, can be delivered at doses low enough to avoid meaningful weight loss. The Hendershot trial (a significant randomized controlled trial testing semaglutide for alcohol use disorder) used low-dose semaglutide, and the effect on craving was still substantial. This is a hypothesis that urgently needs testing in underweight addiction populations specifically. It also points to where the broader HOT framework, outlined in detail in Rethinking Health, becomes relevant.
Myostatin and activin inhibitors, the muscle-building drugs discussed in the Rethinking Health report could, in principle, counteract the physical depletion that addiction causes, rebuilding lean mass even in a state of caloric restriction. A low-dose GLP-1 to quiet the craving, paired with a myostatin blocker to rebuild the body. No such combination has been trialled for addiction yet, but this could be the key to addressing addiction issues in the most vulnerable population.
Mass adoption
The therapeutic landscape for addiction is a patchwork that has been built substance-by-substance, requiring separate diagnoses, separate prescriptions, separate treatment pathways for each individual addiction. GLP-1 agonists appear to operate at the level of the the reward evaluation system itself rather than at the level of any individual substance’s receptor which means that a single therapy may reduce craving and compulsive behavior across multiple substances and addictive behaviors simultaneously.
The criminal justice implications alone are significant. Alcohol is implicated in roughly 40% of all violent crimes and a similar proportion of traffic fatalities and domestic abuse cases. Close to 65% of the prison population has an active substance use disorder. Reducing the grip of addiction to all the most popular drugs at population scale would ripple through courts, prisons, emergency rooms, and families in ways that are difficult to fully quantify but impossible to overstate.
There is also a profound implication for prevention. Existing addiction medicines are treatments for established disorders. They require a diagnosis, a clinical encounter, a prescription, and sustained engagement with a healthcare system that many addicted individuals are reluctant or unable to access. GLP-1 agonists, by contrast, are already being prescribed for obesity and metabolic health to tens of millions of people, and their anti-addiction effects appear to occur as a natural consequence. The person who starts semaglutide to address their weight may find, without seeking it, that their drinking diminishes, their smoking becomes easier to resist, and their relationship with compulsive behaviors changes. This is a major difference from anything that has existed before. Not treatment-first, but optimization-first, with treatment as a byproduct.
The uncertainties that remain for the direct application of GLP1s for addiction are real. Whether effects persist when people stop taking the drugs is unknown. Access barriers, cost, stigma, and the reach of the healthcare system will all shape who actually benefits. These are questions for the trials now underway. They are not reasons to wait.
DISCLAIMER: This blog post is for general informational and educational purposes only and does not constitute medical advice. Always seek the guidance of a qualified healthcare provider with any questions about medical conditions or treatments.
A RethinkX Rethinking Health Follow-On Blog | March 2026
This article is part of Rethinking Health, RethinkX’s ongoing series on the Health Optimizer Therapy disruption. RethinkX are experts on global technology disruptions and system change. We are not medical experts. The information in this article is for educational purposes only and does not constitute medical advice.