The discovery that patients taking semaglutide and tirzepatide for obesity or diabetes spontaneously report reduced cravings for alcohol, nicotine, compulsive gambling, and recreational drugs was not predicted by the drugs' original pharmacology rationale. These reports began appearing in patient forums, physician case notes, and the lay press in 2022–2023 — earning headlines like "Ozempic makes people not want to drink" — and generated enormous scientific interest because they suggest that GLP-1 receptor activation has a broad, previously uncharacterized effect on the dopamine reward system, not just on the hypothalamic appetite circuits that were the original target.
The mechanistic basis for this observation had actually been established in rodent research a decade earlier, primarily by the laboratory of Elisabet Jerlhag at the University of Gothenburg. Jerlhag's group demonstrated from 2006 onward that GLP-1 receptors are expressed in the mesolimbic reward pathway — the brain circuit that drives motivation, pleasure, and addiction — and that activating these receptors suppresses the dopamine release that addiction depends on. The human reports of spontaneous addiction-like behavior reduction on GLP-1 agonists are consistent with exactly this mechanism playing out in patients at clinical doses.
Mesolimbic GLP-1R Biology
the neuroscience of why GLP-1 suppresses reward: THE MESOLIMBIC DOPAMINE PATHWAY: the brain's primary reward circuit; consists of: dopamine neurons in the VTA (ventral tegmental area, in the midbrain) → project axons to: nucleus accumbens (NAc, primary "reward hub"), prefrontal cortex (PFC, decision-making), amygdala (emotional memory); when a rewarding stimulus occurs (eating palatable food, consuming alcohol, using drugs, gambling win): VTA dopamine neurons fire → dopamine released in NAc → experienced as pleasure/reward → drives motivation to repeat the behavior; addiction is defined partly as pathological sensitization of this system — repeated substance exposure changes the VTA-NAc circuit, increasing craving and compulsion; GLP-1 RECEPTORS IN THE REWARD CIRCUIT: GLP-1Rs are expressed in: NAc shell (the most reward-sensitive subregion); VTA dopamine neurons themselves; lateral hypothalamus (LH) → a key input to the VTA; brainstem nucleus tractus solitarius (NTS) → sends GLP-1 signals to VTA via ascending projections; GLP-1R IN THE NAC — JERLHAG RESEARCH (2006–2022): Jerlhag E et al. (2006, Neuropharmacology): intra-NAc injection of exendin-4 (a potent GLP-1R agonist, the drug exenatide) in rodents: significantly reduced alcohol-induced dopamine release in the NAc (measured by in vivo microdialysis); Jerlhag 2009 (Addiction Biology): systemic GLP-1R agonist (exendin-4) reduced voluntary alcohol intake in Wistar rats by 40–60%; the effect was dose-dependent and reversible; Suchankova P et al. (Addiction Biology 2015): GLP-1R agonist (liraglutide) reduced alcohol preference in rats bred for high alcohol preference (AA rats) — suggesting the effect operates in models of elevated vulnerability to AUD; Egecioglu E et al. (2013, Psychopharmacology): GLP-1R agonism reduced cocaine-induced dopamine release in NAc AND reduced voluntary cocaine intake in rodents → extending the mechanism beyond alcohol; Shirazi RH et al. (2013, Addiction Biology): amphetamine-induced locomotor sensitization (a model of addiction neuroplasticity) was attenuated by GLP-1R agonism; MECHANISM IN NAC: GLP-1R activation in NAc neurons: (1) directly inhibits adenylyl cyclase (GLP-1R is a Gs-coupled GPCR but GLP-1R in neurons may signal via Gi/o in some contexts); (2) reduces cAMP → reduces PKA activity → modulates AMPA/NMDA receptor phosphorylation → reduces excitatory synaptic transmission at dopamine-responsive neurons; (3) activates KATP channels → hyperpolarizes dopamine neurons in VTA → reduces their firing rate → less dopamine released in NAc per rewarding event; FOOD NOISE vs SUBSTANCE CRAVING: the same mesolimbic circuits drive both hedonic eating and substance addiction; GLP-1R agonism suppresses BOTH types of reward-driven behavior through the shared NAc/VTA mechanism; this explains why patients report simultaneous reduction in food cravings AND substance cravings without separately targeting either
Hendershot 2024 — The Key RCT
the first controlled trial in humans with AUD: HENDERSHOT CS ET AL. (2024, Alcoholism: Clinical and Experimental Research): DESIGN: double-blind, placebo-controlled, randomized pilot RCT; POPULATION: N=48 adults (non-treatment-seeking heavy drinkers; met criteria for alcohol use disorder per DSM-5; BMI ≥25; not currently seeking AUD treatment); INTERVENTION: semaglutide 0.25mg week 1 → 0.5mg weeks 2–4 → 1.0mg weeks 5–9 (standard dose escalation) vs placebo SC injection weekly; DURATION: 9 weeks total; PRIMARY OUTCOME: Penn Alcohol Craving Scale (PACS) — a validated 5-item self-report craving measure; SECONDARY OUTCOMES: weekly alcohol consumption (drinks per week via timeline follow-back method); heavy drinking days (≥5 drinks per occasion for men, ≥4 for women); percent drinking days; RESULTS: PACS CRAVING SCORE: semaglutide: significant reduction from baseline vs placebo (effect size d ≈ 0.7, p < 0.05 — approximately −40% craving reduction vs placebo); ALCOHOL CONSUMPTION (self-reported): semaglutide group: −30% drinks per week vs baseline; placebo group: −14% (regression to mean or therapeutic attention effect); between-group: significant (p < 0.05); HEAVY DRINKING DAYS: reduction in semaglutide group reached trend significance; WEIGHT LOSS (expected co-effect): semaglutide group lost ~3.5kg vs baseline vs minimal change placebo; ADVERSE EVENTS: nausea: 38% semaglutide vs 8% placebo (expected for GLP-1); no serious adverse events; INTERPRETATION: this is a PILOT trial — the N=48 sample generates hypothesis-confirming signal but is underpowered for definitive clinical recommendations; larger Phase 2 trials are needed; the effect size (d≈0.7) is comparable to naltrexone for alcohol craving; PRIOR CASE SERIES: Klausen MK et al. (2022, JCI Insight): liraglutide in patients with AUD + obesity: significant reduction in alcohol consumption and craving vs historical controls; case series data; Probst CC et al. (2023, NEJM Evidence): letter describing 3 cases of dramatic spontaneous alcohol intake reduction on semaglutide; the anecdotal groundswell preceding RCT data
Beyond Alcohol — Nicotine, Opioids, Compulsive Behavior
survey evidence and emerging RCTs across addiction types: NICOTINE / SMOKING CESSATION: SURVEY DATA: multiple surveys of GLP-1 agonist users (e.g., KFF 2023, various Twitter/Reddit analyses) report that a substantial minority of users spontaneously reduced or quit smoking without specifically trying; the effect is described as "I just didn't feel like smoking" — paralleling the food noise reduction for food cravings; ONGOING RCTS: as of mid-2026, multiple Phase 2 RCTs are enrolling to test GLP-1R agonists in nicotine use disorder; early results expected 2025–2027; THEORETICAL MECHANISM: nicotine addiction is also mesolimbic-dopamine dependent; GLP-1R agonism in VTA/NAc would be expected to reduce nicotine-stimulated dopamine release by the same mechanism as alcohol; GLP-1R agonism in rodents: Egecioglu 2013 (PLoS ONE): liraglutide significantly reduced nicotine-induced dopamine release in NAc; reduced voluntary nicotine self-administration; OPIOID USE DISORDER: PRECLINICAL: Bornebusch AB et al. (2019, PLoS ONE): GLP-1R agonism reduced morphine-induced dopamine release in NAc and reduced morphine-conditioned place preference (a rodent addiction model); HUMAN: case reports of reduced opioid craving in patients on semaglutide; no large RCT completed as of mid-2026; COMPULSIVE BEHAVIORS (NON-SUBSTANCE ADDICTIONS): gambling, compulsive shopping, binge eating disorder, sex addiction: these share mesolimbic reward pathway activation; numerous patient-reported anecdotes of simultaneous reduction in compulsive shopping, gambling urges, and even social media compulsion on GLP-1 agonists; these are consistent with mesolimbic GLP-1R suppression reducing the dopamine burst that reinforces any compulsive behavior; BINGE EATING DISORDER (BED): semaglutide is being studied in BED separately from obesity RCTs; a 2023 NEJM Evidence letter: dramatic binge eating reduction in 3 patients with BED on semaglutide that was not explained by weight loss alone; GLP-1 × BEHAVIORAL ADDICTION CLINICAL FRAMEWORK: the emerging model is that GLP-1R agonism represents a general mesolimbic "tone reducer" — it does not block specific receptors like naltrexone (opioid receptor antagonist for AUD) but broadly reduces the reward salience of any stimulus-driven behavior by reducing NAc dopamine availability
Clinical Context and Limitations
what this means for practice and what we don't know yet: CURRENT REGULATORY STATUS (mid-2026): NO GLP-1 agonist is FDA-approved for ANY addiction or substance use disorder indication; approved indications remain: T2DM (Ozempic/Mounjaro), obesity (Wegovy/Zepbound), CV risk reduction in T2DM (Ozempic); off-label prescribing for addiction is occurring (particularly by addiction medicine physicians for AUD, given Hendershot 2024 pilot data), but without a formal indication; COMPARISON TO APPROVED AUD TREATMENTS: naltrexone (Vivitrol, Revia): FDA-approved for AUD; NNT ~12 for significant drinking reduction; mechanism: opioid receptor antagonism in VTA → reduces alcohol-stimulated dopamine release; acamprosate (Campral): FDA-approved; mechanism: glutamate/GABA modulation; reduces alcohol withdrawal craving; disulfiram (Antabuse): aversion therapy; not dopamine-pathway; HOW GLP-1 COMPARES TO NALTREXONE: mechanism: different receptors (GLP-1R vs opioid receptor) but same downstream target (NAc dopamine); naltrexone is best-evidence for AUD relapse prevention; semaglutide's effect size in Hendershot 2024 is COMPARABLE to naltrexone in similar pilot studies; potential advantage of GLP-1 for AUD patients with co-occurring obesity/metabolic syndrome: single drug addresses both conditions; WHAT WE DON'T KNOW: (1) Does weight loss CAUSE the AUD improvement (less alcohol with caloric restriction) or does GLP-1R agonism independently reduce cravings? Hendershot 2024 tried to separate these but pilot data is insufficient; (2) Do GLP-1 agonists prevent relapse in abstinent AUD patients (as naltrexone does) or only reduce use in active drinkers? (3) Long-term maintenance data (most AUD pharmacotherapy studies are 6–12 months; GLP-1 effects may require continued dosing; what happens on cessation?); (4) Optimal dose for AUD — is the weight-management dose (2.4mg semaglutide) better than the T2DM dose (1mg)? (5) Tirzepatide (dual GIP+GLP-1): the GIP receptor is ALSO expressed in the NAc — does dual agonism produce additive anti-addiction effects? No data yet; PHASE 3 TRIALS NEEDED: large, sufficiently-powered, multi-center RCTs with AUD as primary endpoint are required before this becomes standard of care; several are actively enrolling as of 2026
| Drug | Mechanism | AUD Evidence Level | Weight Effect | Status |
|---|
| Naltrexone (Vivitrol) | Opioid receptor antagonist → ↓ dopamine in NAc | FDA-approved; NNT ~12 | Mild weight loss | Standard of care |
| Acamprosate (Campral) | Glutamate/GABA modulation → ↓ withdrawal craving | FDA-approved; best in abstinent patients | Neutral | Standard of care |
| Disulfiram (Antabuse) | Aldehyde dehydrogenase inhibition → aversion | FDA-approved; poor adherence | Neutral | Rarely used |
| Semaglutide (off-label) | GLP-1R in NAc/VTA → ↓ dopamine reward signal | Pilot RCT (Hendershot 2024); promising effect size | Significant weight loss | Off-label; Phase 3 needed |
| Liraglutide (off-label) | Same as semaglutide; shorter half-life | Case series (Klausen 2022) | Moderate weight loss | Off-label; limited human data |
| Tirzepatide (off-label) | GIP+GLP-1R; GIP also in NAc | No human AUD RCT yet | Greatest weight loss | Theoretical; trials pending |
Clinical Considerations — GLP-1 for Patients with AUD + Obesity (Physician Context)
CURRENT EVIDENCE FRAMEWORK: Hendershot 2024 (N=48, pilot) shows meaningful signal for semaglutide in heavy drinkers — the ~40% craving reduction and −30% drinks/week are clinically meaningful effects; effect size comparable to naltrexone; but pilot data from a non-treatment-seeking population requires replication in a powered Phase 3 trial before this can be a first-line AUD recommendation; OFF-LABEL CLINICAL SCENARIO — AUD + OBESITY: for patients who have BOTH AUD and obesity/T2DM: semaglutide or tirzepatide may be uniquely positioned to address both simultaneously; in this specific patient population, an addiction medicine physician or addiction psychiatrist familiar with the evidence may consider GLP-1 agonist as part of a comprehensive AUD treatment plan (combined with behavioral therapy, possibly alongside naltrexone for additional coverage); not a replacement for naltrexone in patients with AUD-only (no metabolic indication) until Phase 3 data is available; KEY INTERACTIONS IN AUD PATIENTS STARTING GLP-1: GLP-1 agonists slow gastric emptying → slows alcohol absorption → higher peak BAC for a given drink (if patients continue to drink, they may experience greater intoxication per drink early in treatment while gastric emptying is maximally slowed); warn patients: drink very slowly if drinking at all during GLP-1 initiation; nausea + alcohol = vomiting risk; ALCOHOL ON AN EMPTY STOMACH (GLP-1 users often have delayed gastric emptying with poor appetite): alcohol absorption from the small intestine (not stomach) is not significantly delayed by GLP-1 — but reduced food intake means alcohol hits a less-buffered stomach → subjectively faster effect; patients may feel more intoxicated than expected on their usual amount; WHAT PATIENTS ON SEMAGLUTIDE REPORT (qualitative pattern): the most consistent self-report is a general reduction in "wanting" — a decrease in the pull toward reward-seeking behaviors; food, alcohol, and substances are reported as "less interesting" rather than actively aversive (unlike disulfiram which causes illness); the experience is described as the reward salience of the stimulus being reduced: "I can take it or leave it" vs "I need it" — consistent with reduced NAc dopamine tone; MONITORING: liver function (alcohol + GLP-1; alcohol is hepatotoxic, GLP-1 may be hepatoprotective via MASH mechanism); no specific AUD-related monitoring beyond standard GLP-1 monitoring (pancreatitis symptoms, thyroid); alcohol withdrawal risk: patients who significantly reduce heavy alcohol use on GLP-1 may experience withdrawal — assess CIWA-Ar score if rapid drinking reduction occurs; alcohol withdrawal can be life-threatening (Wernicke's encephalopathy risk, seizures) and may require supervised medical detox.