GLP-1 Drugs and Alcohol Use Disorder: What the Trial Evidence Actually Shows in 2026

Patient-reported and case-series accounts of semaglutide reducing alcohol cravings have made the rounds on TikTok and in the lay press for two years. The trial-level evidence is much narrower than the social-media coverage suggests. As of early 2026, exactly one completed randomized controlled trial of semaglutide for alcohol use disorder has been published (Hendershot et al., JAMA Psychiatry, February 2025, n=48) [1]. That trial reported a statistically significant reduction in drinks per drinking day over 9 weeks but did not show a difference in the number of drinking days. There is real preclinical animal evidence and a plausible mesolimbic dopamine mechanism. There is also one earlier negative RCT of a different GLP-1 agonist (exenatide) for the same indication [2]. This article walks through the actual published evidence, the mechanism, and where semaglutide for AUD honestly stands relative to the three FDA-approved alcohol use disorder medications.

What is actually in the published literature

The trial that everyone is citing — and the only Phase 2 RCT of semaglutide for AUD that has been published in a peer-reviewed journal as of this writing — is Hendershot et al., published in JAMA Psychiatry on February 12, 2025 [1]. Here are the verified details:

• Sample size: 48 adults with alcohol use disorder (DSM-5 diagnosis), randomized 1:1 to semaglutide or placebo
• Duration: 9 weeks of active treatment
• Dose: Semaglutide titrated weekly from 0.25 mg to 1.0 mg subcutaneously — notably, this is lower than the 2.4 mg weight-loss dose used in STEP trials and the 1.0-2.0 mg diabetes doses used in SUSTAIN
• Setting: Outpatient laboratory protocol with a controlled alcohol self-administration session
• Primary outcome: Change in alcohol consumption during a laboratory drinking paradigm

The headline result was that semaglutide significantly reduced the amount of alcohol consumed per drinking day during the laboratory session (β = −0.48 grams per drinking day, P=0.01) [1]. The trial also reported reductions in alcohol craving and weekly alcohol consumption in the semaglutide arm relative to placebo. Importantly, the trial did not demonstrate a statistically significant reduction in the number of drinking days — the effect was on intensity of drinking when drinking occurred, not on abstinence. The trial was not powered or designed to evaluate sustained abstinence as an outcome.

Two pieces of additional context matter for interpreting this result responsibly. First, with n=48 the trial is small by RCT standards and was framed by the authors themselves as a Phase 2 proof-of-concept study, not a definitive efficacy trial. Second, the dose used (1.0 mg) is below the typical weight-loss dose, which means the trial doesn't directly answer what the 2.4 mg Wegovy dose would do for AUD outcomes — it could be more effective, less effective, or have a different side-effect profile.

The earlier negative trial: exenatide in 2022

Before the Hendershot trial, the largest randomized GLP-1 trial in AUD was the Klausen et al. exenatide study published in JCI Insight in 2022 [2]. 127 participants with alcohol use disorder were randomized to once-weekly exenatide (a different GLP-1 agonist) or placebo for 26 weeks. The headline result was negative: exenatide did not reduce heavy drinking days compared with placebo over the 26-week trial. However, in a prespecified subgroup analysis, participants with a BMI greater than 30 showed reduced heavy drinking days on exenatide. Functional MRI showed reduced alcohol cue reactivity in the ventral striatum and septal area in the exenatide arm.

The honest summary of the human GLP-1 + AUD evidence is therefore: one negative larger trial of an older GLP-1 (exenatide) with a positive obese-subgroup signal, and one positive smaller trial of semaglutide. That is the entirety of the randomized human evidence as of early 2026. Anything else you read about semaglutide and alcohol cravings is either preclinical, an observational study, a case report, or anecdote.

The preclinical animal evidence and the proposed mechanism

The preclinical case for GLP-1 agonists in addiction is substantial. Aranäs et al. (EBioMedicine, 2023) showed that semaglutide reduced alcohol intake and relapse-like drinking in both male and female rats across multiple exposure paradigms [3]. Earlier work from Vallöf and colleagues established that liraglutide attenuated the reinforcing properties of alcohol in rodent models [4]. The Jerlhag (Frontiers in Pharmacology, 2023) review summarized the broader preclinical literature across multiple GLP-1 agents and multiple addictive substances [5].

The proposed mechanism is that GLP-1 receptors are expressed in the mesolimbic dopamine pathway — specifically in the ventral tegmental area and nucleus accumbens, the brain's primary reward circuit. GLP-1 receptor activation appears to attenuate dopamine release in response to addictive stimuli (alcohol, nicotine, cocaine, opioids) in animal models. If this mechanism translates to humans, you'd expect a general reward-system dampening that affects multiple substances of abuse, not just food. The Quddos et al. 2023 Scientific Reports analysis of patients with obesity on semaglutide and tirzepatide reported reduced alcohol consumption as a self-reported secondary observation [6], which is consistent with this mechanism but does not constitute trial-level evidence.

What the observational data actually shows

Beyond the two RCTs, the most-cited observational evidence comes from Wang et al. (Nature Communications, 2024) [7], a large EHR-based analysis comparing semaglutide-treated patients with controls treated with other anti-obesity or anti-diabetes medications. The study reported a reduced incidence and recurrence of alcohol use disorder diagnoses in the semaglutide arm. This is a real and intriguing signal, but it has the well-known limitations of all retrospective EHR cohort analyses: confounding by indication, residual selection bias, differential follow-up between arms, and the fundamental problem that an “AUD diagnosis” in an EHR is a proxy for what a clinician chose to document, not for actual drinking behavior. EHR cohort findings of this kind generate hypotheses; they don't establish efficacy.

The standard of care: what the FDA-approved AUD medications actually do

Semaglutide for AUD is being discussed in the context of an existing standard of care that most patients have never been offered. The Bahji et al. 2022 systematic review and network meta-analysis of pharmacotherapies for alcohol use disorder (Journal of Addiction Medicine) [8] is the best recent summary. The three FDA-approved AUD medications and their approximate effect sizes:

All three are off-patent, inexpensive, and dramatically underprescribed. The Bahji meta-analysis estimated that fewer than 10% of US patients with AUD receive any FDA-approved medication for it [8]. The clinical significance of the GLP-1 + AUD story isn't that it solves a problem with no existing solution — it's that the GLP-1 mechanism (reward circuit dampening) is genuinely different from any of the three approved drugs and may help patients who don't respond to or tolerate them.

The safety questions specific to AUD patients

Patients with active heavy alcohol use are not the population in whom semaglutide has been studied. The STEP and SUSTAIN trials excluded patients with significant alcohol use, and the published Phase 2 AUD trials are small. A few specific safety considerations:

1. Pancreatitis. Heavy alcohol use is one of the leading causes of acute and chronic pancreatitis. GLP-1 agonists carry a label warning for pancreatitis. The combination is at least theoretically additive, and the published trials are not large enough to characterize this risk.
2. Hypoglycemia interaction. Alcohol blunts hepatic gluconeogenesis. GLP-1 agonists enhance insulin secretion in a glucose-dependent manner. The combination can produce more pronounced hypoglycemia than either alone, particularly in fasted or undernourished patients with chronic alcohol use.
3. Nutritional status. Patients with chronic heavy alcohol use frequently have vitamin and mineral deficiencies (thiamine, folate, B12, magnesium). The gastrointestinal side effects of GLP-1 agonists can worsen oral intake and absorption, potentially deepening nutritional deficits in an already vulnerable population.
4. No long-term AUD-specific data. Even the Hendershot trial was 9 weeks. Whether the early laboratory finding holds up over months to years of treatment is completely unstudied.

What this means for patients and prescribers

Honest framing for a patient asking about semaglutide for alcohol cravings in 2026:

• Semaglutide is not FDA-approved for alcohol use disorder. Any prescription for this purpose is off-label.
• The trial evidence consists of one positive Phase 2 RCT (n=48, 9 weeks, low dose) and one negative larger trial of a different GLP-1 (exenatide).
• The mechanism is biologically plausible and the preclinical animal data is strong, but biological plausibility plus animal data is the same level of evidence that supports dozens of failed addiction drug candidates.
• The three FDA-approved AUD medications (naltrexone, acamprosate, disulfiram) are first-line, evidence-based, inexpensive, and almost universally underprescribed. If you're struggling with alcohol use, those should be discussed first with a clinician.
• For patients already taking semaglutide for weight loss who notice reduced alcohol cravings as a side effect, that observation is consistent with what the small trial showed and worth telling your prescriber. It does not, however, mean that semaglutide should replace evidence-based AUD treatment.

The larger Phase 2 SEMALCO trial of semaglutide for alcohol use disorder is underway and will report in the next 1-2 years. If it replicates the Hendershot finding at a larger sample size and longer duration, semaglutide will move meaningfully closer to a serious case for off-label use in carefully selected patients. Until then, the gap between “TikTok says it works” and “trial evidence supports it” remains very wide, and any responsible editorial coverage has to make that gap clear.

Related research

For the broader picture of how semaglutide is being used (and priced) for its FDA-approved indications, see our GLP-1 pricing index and our side-effects investigation. For the cardiovascular outcomes evidence in non-diabetic adults, see our SELECT trial deep-dive. For what happens when patients stop semaglutide, see our STEP-1 / STEP-4 extension review.