GLP-1s and the brain: alcohol, addiction and dementia in 2026

Why anyone thought GLP-1s might work in the brain

GLP-1 receptors are found in the brain as well as the gut and pancreas, including in areas involved in appetite, reward and motivation. GLP-1 drugs reduce appetite partly by acting on these circuits, and patients taking them for diabetes or weight loss began reporting that alcohol, cigarettes or other habits held less appeal. Animal studies had already shown GLP-1 drugs reducing alcohol intake in rodents and primates ¹. At the same time, laboratory work suggested the drugs might protect neurons, and diabetes is a risk factor for dementia, which led to trials in Alzheimer's disease.

Then came very large observational studies. A Veterans Affairs analysis compared 215,970 people with diabetes who started a GLP-1 drug with people starting other diabetes drugs or continuing usual care, across 175 health outcomes. Compared with usual care, GLP-1 use was associated with lower risks of substance use disorders, psychotic disorders and neurocognitive disorders including Alzheimer's disease and dementia, alongside higher risks of gastrointestinal problems, low blood pressure, kidney stones and pancreatitis ². Associations of this kind generate hypotheses. Only randomised trials can test them.

Alcohol: what the trials show

Four randomised, placebo-controlled trials of GLP-1 drugs in alcohol use disorder (AUD) have now been published ^{1 3 4 5}:

The pattern is consistent enough to take seriously. The semaglutide trials, in different populations and with different main outcomes, all found reductions in heavier drinking: fewer drinks on drinking days, fewer heavy drinking days, or less alcohol consumed in a controlled setting ^{3 4 5}. The largest, in 108 people with AUD and obesity, met its primary outcome, with heavy drinking days falling about 14 percentage points more than on placebo ⁵. A 2025 systematic review of the earlier randomised trials reached a similar reading: semaglutide reduced alcohol use, dulaglutide lowered intake in current drinkers, and exenatide had no significant effect on heavy drinking days ⁶.

The cautions are just as important. These trials are small, short (8 to 26 weeks) and mostly phase 2; two missed their primary outcome and found effects only in secondary measures ^{1 4}. The exenatide trial's benefit appeared only in an exploratory subgroup with obesity ¹, and the largest semaglutide trial enrolled only people with obesity ⁵, so whether the effect holds in lean people with AUD is unclear. Larger trials are needed before any GLP-1 drug could be approved for alcohol use disorder, and none is approved today.

How a GLP-1 drug might change drinking

The exenatide trial gave the clearest look at mechanism, even though it missed its main goal. In participants who had brain scans, exenatide reduced the response to alcohol cues in the ventral striatum and septal area, regions central to reward and craving, and dopamine transporter availability was lower than on placebo ¹. The low-dose semaglutide trial found less alcohol drunk and lower peak breath-alcohol in a laboratory session where participants could choose how much to drink, alongside less craving week to week ³. Together these suggest the drugs dampen the pull of alcohol rather than simply making people feel too full or nauseous to drink, though nausea, the commonest side effect, could contribute ⁵.

What the alcohol trials still need to answer

Several practical questions remain open. The trials ran for weeks to six months, so nobody knows whether the effect lasts, or whether drinking rebounds when the drug stops, as weight often does. Most participants were women in one trial and people with obesity in another ^{3 5}, so effects in other groups are uncertain. No trial has compared a GLP-1 drug with naltrexone or acamprosate, the established medicines for AUD, or tested whether combining them helps. And the size of the benefit in the largest trial, about 14 percentage points fewer heavy drinking days than placebo, is meaningful but not dramatic ⁵. Larger and longer phase 3 trials will be needed to answer these questions, and to establish safety in people who are not overweight.

Smoking, cannabis and other substances

The evidence for other substances is thinner still. In the 2025 semaglutide trial, a subgroup of participants who smoked cut their cigarettes per day more than those on placebo ³, and the 2026 oral semaglutide trial found fewer cannabis use days ⁴. Both were secondary findings in small groups, the kind of signal that justifies a dedicated trial rather than a conclusion. The Veterans Affairs cohort's association with lower risk of substance use disorders covers several substances at once and cannot separate cause from the many differences between people prescribed different diabetes drugs ². Trials in cocaine, opioid and nicotine use disorders have been registered, and until they report, claims that GLP-1 drugs treat addiction go beyond the evidence.

Dementia: the trials that said no

Alzheimer's disease is where the observational hope met its sternest test. The evoke and evoke+ phase 3 trials randomised 3,808 people with early symptomatic Alzheimer's disease, average age 72, to oral semaglutide or placebo for 104 weeks. Decline on the Clinical Dementia Rating Sum of Boxes, the main measure, was essentially identical: an increase of 2.3 and 2.2 points on semaglutide against 2.3 and 2.1 on placebo, with no significant difference in either trial ⁷. Adverse events were more common on semaglutide (91.2% against 84.8%) ⁷. The investigators' conclusion was plain: oral semaglutide was not efficacious in slowing clinical progression ⁷.

The liraglutide trial ELAD, in 204 people with mild to moderate Alzheimer's disease without diabetes, also missed its primary outcome, a change in brain glucose metabolism on PET scans. One executive-function measure favoured liraglutide, but daily activities and the overall dementia rating did not differ ⁸. Taken together, these results make it unlikely that GLP-1 drugs slow Alzheimer's disease once symptoms have begun. Whether they might lower the risk of developing dementia, as the cohorts suggest, is a different question that these trials did not test.

It is worth being precise about what evoke tested: oral semaglutide at up to 14 mg a day, flexibly dosed, in people who already had early symptomatic Alzheimer's disease ⁷. It did not test injected semaglutide at weight-loss doses, people without symptoms, or other forms of dementia such as vascular dementia. Those gaps leave room for future trials, but they do not rescue the original hope that these drugs could treat Alzheimer's disease.

Why the cohorts and the trials disagree

The gap between the observational studies and the Alzheimer's trials is a lesson in reading evidence. People prescribed GLP-1 drugs differ from people prescribed other diabetes drugs in many ways, including health, income, access to care and how closely they are monitored, and even sophisticated statistical adjustment cannot remove all of those differences. The drugs also cause weight loss and better blood sugar control, which could lower dementia risk without any direct effect on the brain. And a cohort that looks at risk of a first diagnosis asks a different question from a trial that treats people who already have symptoms.

None of this makes the cohort findings worthless; they point to where trials should look. It means a headline that a drug 'cuts dementia risk' based on records data should be read as 'is associated with', until a randomised trial says otherwise. Our guide to reading a study explains why the rung of the evidence ladder matters.

Mood and psychiatric safety

The brain effects of GLP-1 drugs are a safety question as well as a hope. The Wegovy and Zepbound labels used to carry a warning about suicidal behaviour and ideation, inherited from older weight-loss drugs; both labels list that warning as removed in February 2026 ^{9 10}. The Veterans Affairs atlas found lower, not higher, risks of several psychiatric outcomes with GLP-1 use, while also listing the drugs' established risks ². Removal of a label warning is not a guarantee for any individual: anyone who notices new or worsening low mood, anxiety or suicidal thoughts on any medicine should contact their prescriber promptly.

What this means now

For alcohol, the evidence is the most encouraging of any brain-related use, but it is early. People with alcohol problems have approved treatments, such as naltrexone and acamprosate, and support that works; a GLP-1 drug should not be started for drinking outside a trial or a careful discussion with a doctor. For people already taking a GLP-1 drug for diabetes or weight, any reduction in drinking is a welcome side effect, not a treatment plan.

For dementia, the large trials have answered the main question: semaglutide does not slow early Alzheimer's disease ⁷. For other substances, the data are signals. As of October 3, 2026, no GLP-1 drug is approved for any addiction or for dementia ¹¹, and research or compounded versions sold for these purposes carry all the usual risks of unregulated products. Our GLP-1 transition hub covers the approved uses.

Frequently asked questions

Does Ozempic reduce alcohol cravings?

Small randomised trials of semaglutide in alcohol use disorder found reductions in heavy drinking, drinks per drinking day or craving, but they were short and phase 2. No GLP-1 drug is approved for alcohol use disorder.

Does semaglutide prevent dementia?

It has not been shown to. The evoke and evoke+ trials found oral semaglutide did not slow decline in 3,808 people with early Alzheimer's disease. Observational studies link GLP-1 use with lower dementia risk, but that is an association, not proof.

Can GLP-1 drugs treat addiction?

Not yet. The evidence for alcohol is promising but early, and for smoking, cannabis and other drugs it comes from secondary findings in small trials and observational studies. Larger trials are under way.

Why did the Alzheimer's trials fail when cohort studies looked good?

Cohorts compare people who differ in many ways besides the drug, and they ask about developing dementia rather than slowing it once symptoms start. Randomised trials remove those differences, and in Alzheimer's disease they found no benefit.

Related on Grey Peptides

• Semaglutide
• Exenatide
• GLP-1 transition hub

Sources

1. Klausen, M. K., et al. (2022). Exenatide once weekly for alcohol use disorder investigated in a randomized, placebo-controlled clinical trial. JCI Insight, 7(19). PMID: 36066977
2. Xie, Y., et al. (2025). Mapping the effectiveness and risks of GLP-1 receptor agonists. Nat Med, 31(3), 951-962. PMID: 39833406
3. Hendershot, C. S., et al. (2025). Once-Weekly Semaglutide in Adults With Alcohol Use Disorder: A Randomized Clinical Trial. JAMA Psychiatry, 82(4), 395-405. PMID: 39937469
4. Schacht, J. P., et al. (2026). Oral Semaglutide for Alcohol Use Disorder: A Randomized Clinical Trial. Am J Psychiatry, 183(9), 636-645. PMID: 42522065
5. Klausen, M. K., et al. (2026). Once-weekly semaglutide versus placebo in patients with alcohol use disorder and comorbid obesity: a randomised, double-blind, placebo-controlled trial. Lancet, 407(10540), 1687-1698. PMID: 42070571
6. Patel, S., et al. (2025). GLP-1 receptor agonists and alcohol use disorder: a systematic review. Alcohol Alcohol, 61(1). PMID: 41273789
7. Cummings, J. L., et al. (2026). Efficacy and safety of oral semaglutide 14 mg (flexible dose) in early-stage symptomatic Alzheimer's disease (evoke and evoke+): two phase 3, randomised, placebo-controlled trials. Lancet, 407(10544), 2167-2179. PMID: 41865758
8. Edison, P., et al. (2026). Liraglutide in mild to moderate Alzheimer's disease: a phase 2b clinical trial. Nat Med, 32(1), 353-361. PMID: 41326666
9. Novo Nordisk. Wegovy (semaglutide) US prescribing information, Recent Major Changes (Suicidal Behavior and Ideation warning removed 02/2026) (DailyMed set ee06186f, effective June 18, 2026); read in full October 3, 2026.
10. Eli Lilly. Zepbound (tirzepatide) US prescribing information, Recent Major Changes (Suicidal Behavior and Ideation warning removed 02/2026) (DailyMed set 487cd7e7, effective August 28, 2026); read in full October 3, 2026.
11. Grey Peptides encyclopedia: approved indications for semaglutide, liraglutide, exenatide and tirzepatide, from their labels and Drugs@FDA. Read October 3, 2026.

Educational information, not medical advice. Each dose in this guide names its source, an approved label or a published study. None is a recommendation for you. An unapproved compound has no established safe or effective human dose, and products sold for “research use only” are not made or tested for people. Talk to a doctor before acting on anything on this site, including before you start, stop or change any medicine or dose.