The first time it came up in the practice was several years ago. A long-standing patient with alcohol use disorder, a woman in her early fifties who had done the work seriously and who had been stable for some time, mentioned in passing, near the end of an otherwise routine appointment, that the cravings had quieted since she started semaglutide for weight loss. She said it the way one mentions a small change in sleep — not as a clinical event, just as something noticed. The cravings, she said, were not gone. They were further away. She did not know what to make of it. Neither, in that moment, did the physician sitting across from her.
In the years since, the same conversation has happened in clinics across the country. Almost every working addiction physician has had a version of it. The patient names a different substance — alcohol, nicotine, food, stimulants, in some accounts opioids — and a different GLP-1 receptor agonist — semaglutide, tirzepatide, liraglutide — and a different incidental reason for the prescription. The reports are now common enough that they have crossed the threshold from anecdote into something the field has to explain.
The public conversation about GLP-1s and addiction has, so far, been split between two unsatisfying framings. The first is the miracle framing — the medication ends cravings, the disease is solved, the prescriptions should be written. The second is the threat framing — the medications are being used off-label without adequate evidence, the marketing is ahead of the science, the prescribing should be slowed. Both framings are arguing about whether GLP-1s are an addiction treatment. That is the wrong question.
The right question is what the medications are showing us about addiction itself.
A brief sketch of the mechanism, kept honest about what is known and what is not. GLP-1 receptors are not confined to the gut, where they were first identified in their role in glucose regulation. They are also distributed throughout the central nervous system, including in the regions involved in reward processing — the ventral tegmental area, the nucleus accumbens, the prefrontal cortex. When GLP-1 receptor agonists cross the blood-brain barrier, as the newer agents do reliably, they appear to modulate dopamine signaling in these regions. The clinical effect — for patients who experience it, and not all do — is a quieting of the reward salience of the substance. The substance does not become aversive. It simply becomes less interesting. The brain stops insisting on it.
This is not the mechanism of any prior addiction medication. Naltrexone blocks the opioid receptor. Acamprosate modulates glutamate. Bupropion affects dopamine and norepinephrine reuptake. Each works on one piece of the addiction picture. GLP-1 receptor agonists appear to be working on a different piece entirely — on the prior question of how reward salience itself is set. They are not extinguishing the response to a substance. They are modulating the regulatory system that decides how loudly the brain will reach for the substance in the first place.
This is the part the field has been slow to say out loud. The crossover between the neurobiology of metabolic regulation and the neurobiology of addiction was not predicted by the framework most clinicians were trained inside. The disease model of addiction, useful as it has been, draws a clean conceptual line between appetite — a regulated drive toward food and drink — and compulsion — a dysregulated drive toward a substance. GLP-1s are showing the field that the line was never as clean as the framework allowed. The same circuitry that decides how hungry the patient feels at noon also decides, in part, how loudly the bottle calls at six. This is not a metaphor. It is a finding that an honest field has to sit with.
Two implications follow, and they pull in opposite directions.
The first implication is conservative. GLP-1s are not, on the current evidence, an addiction treatment. The data are early, the trials are not yet definitive, and the patients for whom the medications quiet cravings appear to be a subset of patients, not all patients. Prescribing a GLP-1 to a patient with substance use disorder, in the absence of a metabolic indication, on the assumption that the medication will function as monotherapy for the addiction, is not justified by what is currently known. A serious clinician does not run ahead of the evidence on a drug that is being prescribed to tens of millions of people for other reasons. The harm of overstating is real. Patients have been told, by people who should know better, that semaglutide cures alcohol use disorder. It does not. Some of those patients have stopped doing the work that was, in fact, treating them, in the belief that the medication had taken over. That is a clinical injury inflicted by enthusiasm.
The second implication is the opposite of conservative. A serious clinician also does not pretend the new evidence is not there. When a long-standing patient reports that the cravings are quieter on a medication being prescribed for an unrelated reason, the right clinical response is curiosity, not dismissal. When a literature builds — as it has now built, in the form of secondary analyses of weight-loss trials, observational studies, and increasingly preregistered clinical trials specifically designed to test the question — the right professional response is to read it, to weigh it, and to update. Ignoring what GLP-1s are showing us about addiction, on the grounds that the medications are not yet for addiction, is a different kind of professional negligence. A field that refuses to revise the framework when the framework cracks is not a serious field.
The clinical question, then, is not should I prescribe a GLP-1 for addiction. The clinical question is what does this patient's response to a GLP-1, prescribed for whatever reason, tell me about the dysregulation I am trying to treat, and how should that information change the rest of the plan. That is a question a careful physician can answer with a particular patient in front of him. It is not a question that resolves at the level of policy or talking point.
There is a second clinical question that follows from the first. Has the patient's relationship to other appetites also shifted? Patients on GLP-1s frequently report not only quieter substance cravings but quieter patterns across the board — less compulsive eating, less compulsive shopping, less compulsive scrolling, less of the low-grade reaching that characterizes the lives of many high-functioning adults under contemporary conditions. Some patients describe this as a relief. Others describe it as a flatness they did not consent to. The field does not yet have a vocabulary for distinguishing the welcome quieting of a pathological reach from the unwelcome quieting of an ordinary one. A serious physician will sit with that distinction patient by patient. The medication is not neutral on this question, and the patient's report of it is the data.
A note for the patient or family reading this who is wondering what to do with it.
If a patient in active recovery is offered a GLP-1 for a metabolic indication, the right next step is a conversation with the physician who is holding the addiction case — not the prescribing physician, not the family member who read about it, not the friend who heard a podcast. The interaction between the medication and the recovery work has to be thought through by someone who knows the patient well enough to predict it. The medication may be helpful. It may be neutral. It may, in some patients, complicate the recovery in ways that are not obvious in the first month. None of these outcomes is foregone. All of them are knowable, in advance, by a physician with continuity on the case.
If a patient is being told by a clinician that a GLP-1 will treat the addiction itself, the right next step is a second opinion. Not because the clinician is necessarily wrong — the data may eventually justify a more direct claim — but because, on the current evidence, the claim is ahead of the field, and a clinician who is ahead of the field on a high-stakes prescription is a clinician whose other judgments deserve scrutiny.
If a patient is being told by a clinician that GLP-1s have nothing to do with addiction and the whole topic is a fad, the right next step is also a second opinion. The phenomenon is real. The framework is updating. A clinician who refuses to engage with new evidence in the field he practices is not the right physician for a patient whose recovery depends on getting the framework right.
The honest position, for now, is the position the field will eventually catch up to. GLP-1s are not the cure for addiction. They are also not nothing. They are, most importantly, a window into a piece of the disease's neurobiology that the prior framework did not adequately describe. A serious practice integrates this. A careless practice oversells it. A defensive practice ignores it. The families who are paying attention can tell the difference, and they should.
The medication is new. The framework is older than it should be. The patients are the same as they always were — sophisticated, attentive, and entitled to a physician whose thinking is at least as current as their own.