Could GLP-1 Drugs Treat Addiction? The Early Evidence

GLP-1 and Addiction

50-56% lower AUD risk with semaglutide

A real-world study of 83,825 patients with obesity found semaglutide was associated with a 50-56% lower risk of both new and recurring alcohol use disorder compared to other anti-obesity medications.

Wang et al., Nature Communications, 2024

Wang et al., Nature Communications, 2024

View as image

People taking semaglutide for weight loss or diabetes started reporting something unexpected: they lost interest in drinking. Some stopped craving cigarettes. A few said even their impulse to gamble faded. These anecdotal reports, first surfacing on social media and patient forums around 2022, triggered a wave of research into whether GLP-1 drugs treat addiction through the same reward-center pathways they use to suppress appetite.

The evidence has moved fast. Between 2023 and 2026, researchers published the first human randomized trial, multiple large observational studies, and dozens of preclinical experiments. The findings are consistent enough to generate genuine scientific excitement, but the research is still in early stages, and the gap between "promising signal" and "proven treatment" remains wide.

The First Human Randomized Trial

The Hendershot et al. trial, published in JAMA Psychiatry in 2025, is the most important piece of evidence so far. This phase 2, double-blind, randomized trial enrolled 48 non-treatment-seeking adults with alcohol use disorder (AUD) at a US academic medical center. Participants received either semaglutide (escalated from 0.25 mg/week to 1.0 mg over 9 weeks) or placebo.^[1]

The primary outcome, alcohol self-administration in a laboratory setting, showed medium-to-large effect sizes favoring semaglutide: grams of alcohol consumed dropped (beta = -0.48, p = 0.01) and peak breath alcohol concentration fell (beta = -0.46, p = 0.03). Semaglutide also reduced drinks per drinking day (beta = -0.41, p = 0.04) and weekly alcohol craving (beta = -0.39, p = 0.01).

Semaglutide did not reduce total drinking days or average drinks per calendar day. The distinction matters: participants drank less per session but did not necessarily drink less often. This pattern, reduced intensity per episode rather than fewer episodes, is different from what naltrexone produces and may reflect a distinct pharmacological mechanism. An incidental finding showed semaglutide also reduced cigarettes per day in current smokers (p = 0.005), supporting the hypothesis that GLP-1 receptor activation dampens reward-seeking across substances.

The trial's limitations are clear. Only 48 participants were enrolled. They were non-treatment-seeking, meaning they were not trying to quit. The maximum dose reached only 1.0 mg (the obesity dose is 2.4 mg). Despite these constraints, it provided the first randomized evidence that a GLP-1 drug can alter alcohol consumption in humans under controlled conditions.

Large Observational Studies

The 83,825-Patient Obesity Cohort

Wang et al. (2024) analyzed electronic health records of 83,825 patients with obesity and found semaglutide was associated with a 50-56% lower risk for both incidence and recurrence of alcohol use disorder compared to other anti-obesity medications over 12 months. The reductions were consistent across gender, age, race, and diabetes status. Similar findings were replicated in a parallel cohort of 598,803 patients with type 2 diabetes.^[2]

The Swedish Nationwide Cohort

Lahteenvuo et al. (2025), also published in JAMA Psychiatry, used a within-individual design across 227,866 Swedish residents with AUD, comparing periods when patients were using GLP-1 agonists versus periods of nonuse. Semaglutide users had a 36% lower risk of AUD hospitalization (aHR 0.64, 95% CI 0.50-0.83). Liraglutide showed a 28% reduction (aHR 0.72). Both drugs outperformed officially approved AUD medications in reducing hospitalization risk.^[3]

The within-individual design is a methodological strength: rather than comparing different people on different drugs, it compared the same person during periods of use versus nonuse, reducing confounding from lifestyle, genetics, and comorbidities.

The US Veterans Study

A 2026 analysis of 606,434 US veterans with type 2 diabetes found GLP-1 receptor agonist use was associated with 14% lower risk of developing any new substance use disorder (alcohol, cannabis, cocaine, nicotine, or opioids), 40% fewer overdoses, and 50% fewer substance-related deaths compared to other diabetes medications. The risk reduction was substance-specific in magnitude: 25% for opioids, 20% for cocaine and nicotine, 18% for alcohol, and 14% for cannabis. After three years, GLP-1 users also had 30% fewer emergency department visits and 25% fewer hospitalizations related to substance use.

The cross-substance consistency in the veterans data is the strongest argument that GLP-1 drugs act on shared addiction biology rather than substance-specific mechanisms. If the effect were driven by weight loss alone, or by improved diabetes control, the reduction would be expected to vary by substance and correlate with metabolic improvement. Instead, the effect was relatively uniform across five different classes of addictive substances.

Early Case Reports and Survey Data

Before the larger studies, Richards et al. (2023) published a case series documenting marked decreases in AUD symptoms in patients receiving semaglutide for weight loss.^[4] Quddos et al. (2023) conducted two studies showing both semaglutide and tirzepatide reduced alcohol consumption in individuals with obesity, providing the first evidence that the dual GLP-1/GIP agonist tirzepatide may share this effect.^[5]

These early reports helped catalyze the formal clinical trials now underway. For a closer look at the alcohol-specific patient experience, including what people on Ozempic report about losing interest in drinking, the pattern is consistent with what the controlled data shows: reduced craving rather than forced abstinence.

The Preclinical Evidence: Beyond Alcohol

Alcohol

The animal data predates and supports the human findings. Aranas et al. (2023) demonstrated that semaglutide reduced both alcohol intake and relapse-like drinking in male and female rats using intermittent-access models.^[6] Chuong et al. (2023) showed semaglutide reduced binge-like drinking in mice and rats and modulated GABA neurotransmission in the central amygdala and infralimbic cortex, two brain regions critical for addiction. Semaglutide increased inhibitory GABA signaling in alcohol-naive animals, suggesting it may prevent the neurochemical changes that drive escalation from casual to compulsive drinking.^[7]

Cocaine

No approved pharmacotherapy exists for cocaine use disorder. Aranas et al. (2025) found semaglutide suppressed cocaine self-administration, reduced the motivation to consume cocaine, blocked cocaine reinstatement (a model of relapse), and attenuated cocaine-evoked dopamine release in the nucleus accumbens. Semaglutide outperformed shorter-acting GLP-1 agonists and did not cause malaise (measured by kaolin intake), suggesting the behavioral changes were not driven by nausea.^[8]

Tirzepatide and Dual Agonism

Edvardsson et al. (2026) tested tirzepatide, the dual GLP-1/GIP receptor agonist, and found it reduced alcohol drinking and relapse-like behaviors in rodents. This suggests the anti-addiction signal may extend beyond pure GLP-1 agonism to the broader incretin system.^[9]

How GLP-1 Drugs Might Reduce Addictive Behavior

GLP-1 receptors are expressed throughout the brain's reward circuitry: the ventral tegmental area (VTA), nucleus accumbens (NAc), amygdala, and prefrontal cortex. Van Bloemendaal et al. (2014) used functional MRI to show that the GLP-1 agonist exenatide reduced food-related brain responses in the insula, amygdala, putamen, and orbitofrontal cortex. These effects were blocked by GLP-1 receptor antagonism, confirming they were mediated specifically through the GLP-1 receptor.^[10]

Edvardsson et al. (2025) identified a specific inhibitory GLP-1 circuit in the lateral septum that modulates reward processing and alcohol intake. GLP-1 receptor activation in this region enhanced GABAergic inhibition, effectively dampening the reward signal that drives compulsive consumption.^[11]

The emerging picture is that GLP-1 drugs do not target any specific substance. Instead, they appear to reduce the salience of reward signals across the board. This explains why the same drug that reduces food intake also reduces alcohol craving, cocaine seeking, and possibly nicotine use. The mechanism is not "anti-alcohol" or "anti-cocaine." It is anti-craving.

This cross-substance effect raises questions about whether GLP-1 agonists might also affect compulsive behaviors beyond substance use, including gambling, compulsive shopping, and binge eating. If the mechanism is truly anti-craving at the circuit level, the implications extend well beyond traditional addiction categories.

Nicotine and Smoking

The evidence for nicotine is thinner but consistent with the broader pattern. The Hendershot RCT's incidental finding that semaglutide reduced cigarettes per day (p = 0.005) was not a primary endpoint, but the effect size was notable in such a small sample. Earlier work by Yammine et al. (2021) tested exenatide combined with nicotine patches in a pilot randomized trial and found the combination may facilitate smoking cessation. Multiple GLP-1 receptor agonists have been shown to reduce nicotine self-administration in rodent models, with the same pattern of dampened dopamine signaling in the nucleus accumbens observed across alcohol, cocaine, and nicotine paradigms.

Opioids

The preclinical data for opioids follows a similar trajectory. Several GLP-1 receptor agonists have reduced self-administration of heroin, fentanyl, and oxycodone in rodent models, and reduced reinstatement of drug seeking after abstinence. An NIDA-sponsored phase 2 clinical trial (NCT protocol published by Freet et al., 2025) is currently evaluating semaglutide for abstinence from illicit opioids in an outpatient population with treatment-refractory opioid use disorder. Given that opioid overdose deaths remain above 80,000 annually in the US, a pharmacotherapy that reduces overdose risk by 40% (as the veterans study suggests) would represent a substantial public health advance if confirmed in a randomized setting.

This mechanism connects to the broader role of GLP-1 and GIP as incretin hormones that modulate far more than blood sugar. The gut-brain axis they operate through influences reward, motivation, stress, and satiety in ways that were not anticipated when these drugs were developed for diabetes.

What Remains Unknown

The research trajectory is compelling, but the limitations are substantial:

No phase 3 trial exists. The only human RCT enrolled 48 participants over 9 weeks. Multiple phase 2 and phase 3 trials are now underway, including NIDA-sponsored studies on opioid use disorder (NCT protocol published by Freet et al., 2025), but none have reported results. Until at least one adequately powered trial demonstrates efficacy, the evidence remains preliminary.

Observational studies cannot prove causation. The Wang and Lahteenvuo studies show associations, not effects. Patients who are prescribed GLP-1 drugs may differ systematically from those who are not. The within-individual design of the Swedish study reduces but does not eliminate confounding.

Dose, timing, and duration are unknown. The Hendershot trial used a low dose (up to 1.0 mg/week). Whether higher doses produce stronger effects, whether the benefits persist after stopping the drug, and how long treatment would need to continue are all open questions.

The nausea problem. GLP-1 drugs cause nausea in 20-40% of users during dose escalation. In addiction populations, nausea could be misinterpreted as a sign of withdrawal or could reduce medication adherence. The Aranas cocaine study specifically tested for malaise and found none, but this has not been systematically studied in human addiction populations.

No head-to-head comparison with existing treatments. Naltrexone, acamprosate, and disulfiram are approved for AUD. No trial has compared a GLP-1 drug against any of these. The Lahteenvuo study's finding that semaglutide outperformed approved AUD medications on hospitalization risk is provocative but comes from observational data, not a controlled comparison.

Off-label prescribing is outpacing evidence. Some clinicians are already prescribing GLP-1 drugs off-label for addiction, driven by patient demand and media coverage. This creates a risk of premature adoption before efficacy, optimal dosing, and safety in addiction populations are established.

GLP-1 drugs carry their own side effects including pancreatitis risk, gallbladder events, and potential thyroid concerns. In patients with substance-related weight changes, the risk-benefit calculus is more complex than in the diabetes population where these drugs have extensive safety data.

The cardiovascular benefit may compound the addiction benefit. GLP-1 drugs have established cardiovascular risk reduction in diabetes and obesity populations. Since alcohol and cocaine use disorder both carry elevated cardiovascular risk, a single drug that reduces both substance use and cardiac events would have outsized clinical value. But this synergy has not been tested prospectively in addiction populations.

Equity and access are unresolved. GLP-1 drugs cost $800-1,300 per month without insurance. Addiction disproportionately affects uninsured and underinsured populations. If GLP-1 drugs prove effective for substance use disorders, the gap between "works in a clinical trial" and "available to the people who need it" could be larger than for any previous addiction therapy. Existing FDA-approved AUD medications like naltrexone cost under $50 per month.

The Bottom Line

The evidence that GLP-1 drugs reduce addictive behavior is consistent across animal models, observational studies, and one small randomized trial. Semaglutide appears to act on shared reward circuitry rather than substance-specific pathways, which explains effects across alcohol, cocaine, and nicotine. No large-scale RCT has confirmed efficacy for any substance use disorder, and no GLP-1 drug is approved for addiction treatment. Multiple trials are underway. The research is genuinely promising, but the standard for clinical adoption has not been met.

Frequently Asked Questions

Can Ozempic help with alcohol addiction?

Semaglutide (the active ingredient in Ozempic and Wegovy) reduced alcohol craving and consumption in a 48-person randomized trial published in JAMA Psychiatry in 2025. A separate study of 83,825 patients found semaglutide was associated with 50-56% lower risk of alcohol use disorder. No phase 3 trial has confirmed these findings, and semaglutide is not FDA-approved for alcohol use disorder.

How do GLP-1 drugs reduce cravings?

GLP-1 receptors are located in brain reward regions including the nucleus accumbens, amygdala, and ventral tegmental area. Activating these receptors appears to dampen the reward signal that drives compulsive consumption. Brain imaging studies show GLP-1 agonists reduce neural responses to rewarding stimuli. A 2025 study identified a specific inhibitory GLP-1 circuit in the lateral septum that modulates reward processing.

Does semaglutide work for cocaine addiction?

In a 2025 preclinical study, semaglutide suppressed cocaine self-administration, reduced motivation to consume cocaine, blocked relapse-like behavior, and lowered cocaine-evoked dopamine levels in rats. No human trial has tested semaglutide for cocaine use disorder. Clinical protocols are being developed, but no results from human studies are available.

Is semaglutide better than naltrexone for alcohol use disorder?

No head-to-head trial has compared semaglutide to naltrexone or any approved AUD medication. A Swedish observational study found semaglutide was associated with greater reductions in AUD hospitalization risk than approved AUD medications, but this was not a controlled comparison. Naltrexone has decades of randomized trial evidence; semaglutide has one small phase 2 trial.

Can GLP-1 drugs help with opioid addiction?

A 2026 analysis of 606,434 US veterans found GLP-1 drug use was associated with 25% lower risk of developing opioid use disorder and 40% fewer overdoses. Animal studies show GLP-1 agonists reduce self-administration of heroin, fentanyl, and oxycodone. An NIDA-sponsored clinical trial is evaluating semaglutide for opioid use disorder, but no results have been published.

Do GLP-1 drugs reduce nicotine cravings?

In the 2025 Hendershot RCT, semaglutide predicted greater reductions in cigarettes per day among participants who smoked (p = 0.005). Earlier pilot work by Yammine et al. (2021) found exenatide combined with nicotine patches may facilitate smoking cessation. The cross-substance effects are consistent with GLP-1 drugs acting on shared reward circuitry rather than substance-specific pathways.