One in eight adults in the US take a glucagon-like peptide-1 receptor agonist (GLP1-RA), and reports of their psychiatric effects have multiplied. The results of early randomized trials now allow us to begin to separate signal from hype.

Depression
There were good reasons to think GLP1-RAs might help depression. In animal studies, they showed anti-inflammatory, neurotrophic, and pro-cognitive effects. However, semaglutide failed to improve mood or executive function in the first randomized trial to test it in major depressive disorder and obesity, despite causing weight loss in that study. In the medical trials, GLP1-RAs improved well-being but not depressive symptoms (Badulescu S et al, Med 2026;7:100916, Hung TH et al, Hum Psychopharmacol 2026;41(3):e70041).

Eating disorders
Among eating disorders, binge eating disorder (BED) is the most promising target, but even here, the evidence is thin and mixed. Liraglutide reduced binge eating in one small controlled trial, but not in another, which was compromised by a pharmacy dispensing error (Robert SA et al, Obes Res Clin Pract 2015;9:301–304; Allison KC et al, Obes Sci Pract 2023;9:127–136). What we don’t know is whether stronger GLP1-RAs like tirzepatide or semaglutide will be more effective, as suggested by a large observational study (Henney AE et al, Diabetes Obes Metab 2026;28:137–150). Since the primary problem in binge eating is not just overeating but loss of control, it may be that lisdexamfetamine (Vyvanse), the only FDA-approved treatment, better targets the core of the condition.

Addictions
As GLP1-RAs took off, media reports suggesting a dampening effect on drug cravings accumulated. At the same time, large observational studies suggested that GLP1-RAs lowered the risk of developing a substance use disorder (SUD) and reduced the overdose risk among those who already had an SUD (Cai M et al, BMJ 2026:392:e086886). The mechanism may involve central mesolimbic reward circuits, where GLP-1 receptors are present, or—in the case of alcohol—slowed gastric emptying, which delays the rise in blood alcohol level.

Controlled trials were initially less impressive. In one, semaglutide reduced alcohol use in a laboratory setting involving non-treatment-seeking subjects with alcohol use disorder (AUD), while more clinically relevant outcomes were mixed. It reduced heavy drinking days, but not drinks per calendar day (Hendershot CS et al, JAMA Psychiatry 2025;82:395–405). Turning to a clinical sample, exenatide failed in a small RCT of AUD, although it did reduce fMRI cue reactivity in reward centers (Klausen MK et al, JCI Insight 2022;7(19):e159863). But a larger 26-week study of semaglutide in 108 individuals with moderate to severe AUD and obesity found that heavy drinking days fell by 41 percentage points in the treatment group compared with 26 percentage points on placebo. The number needed to treat (NNT) for a 2-level WHO risk reduction was 4.3 (Klausen MK et al, Lancet 2026;407:1687–1698).

The positive RCT supports consideration of semaglutide for patients with AUD who meet criteria for GLP1-RAs on obesity grounds. Ongoing trials of stronger GLP1-RAs in amphetamine and opioid use disorders may clarify this signal in other SUDs.

Dementia and Parkinson’s disease
Hopes for GLP1-RAs in dementia have fallen fast. The rationale was sound, with reductions in neuroinflammation, amyloid, and tau in animal studies and positive results in observational studies. But both semaglutide and liraglutide failed to slow dementia in three large, industry-sponsored trials of early Alzheimer's disease. Whether they prevent dementia is unknown (Edison P et al, Nat Med 2026;32:353–361; Cummings JL et al, Lancet 2026;S0140-6736(26)00459-9). In Parkinson’s disease, the evidence is mixed: Early phase II trials showed promise, but a larger phase III trial found no improvement in motor symptoms (Vijiaratnam N et al, Lancet 2025;405(10479):627–636).

Antipsychotic weight gain
This is where GLP1-RAs have delivered, especially for patients on clozapine or olanzapine, earning them a thumbs up in new schizophrenia practice guidelines (McCutcheon RA et al, Lancet Psychiatry 2025;12:384–394). In the COaST study, 36 weeks of semaglutide led to a 14% body weight change, with no impact on clozapine levels (Siskind D et al, Lancet Psychiatry 2025;12:493–503). A study of semaglutide for patients with schizophrenia spectrum disorder on olanzapine or clozapine with early metabolic abnormalities (HbA1c 5.4%–7.4%) extended this evidence. There was about 20 lb more weight loss at 26 weeks with treatment than placebo, and an NNT of about 3 for achieving a low-risk HbA1c (< 5.4%) (Sass MR et al, JAMA Psychiatry 2026;83(2):128–138). GLP1-RAs did not improve cognition or psychosis in these trials.

Bariatric surgery is associated with increased suicide risk, so GLP1-RAs could be a safer option for psychiatric patients with obesity, and this use is on-label. The FDA recommends GLP1-RAs in those with BMI ≥ 30 kg/m² or BMI ≥ 27 kg/m² plus a weight-related medical disorder (prediabetes/diabetes, hypertension, dyslipidemia, central adiposity/metabolic syndrome). While psychiatrists may prescribe them for obesity, I would defer to the medical provider if the patient has diabetes.

New psychiatric risks
The safety story has become clearer in the last few years. We now have data from RCTs in dementia, eating disorders, mood disorders, and schizophrenia where these medications raised no new safety concerns, leading the FDA to remove the suicidality warning in January 2026. Anecdotally, there are reports of anhedonia and emotional blunting on GLP1-RAs. This isn’t found in the clinical trials but is plausible as GLP-1 receptors are expressed in the mesolimbic reward system.

Still, psychiatrists should heed a few practical warnings. First, there are six case reports of lithium toxicity after starting a GLP1-RA, particularly semaglutide (Al-Soleiti M et al, J Clin Psychopharmacol 2025;45:613–618). Most cases occurred in the first two weeks of starting the GLP1-RA, so it’s a good idea to check lithium in that period. The other risk is slowed gastric motility. This is particularly relevant for patients already on meds that slow down the GI tract, like clozapine and anticholinergics. Monitor bowel frequency more closely when GLP1-RAs are started for these patients.

CARLAT VERDICT
GLP1-RAs are ready to be used for obesity associated with antipsychotics, but it is best to defer to the medical provider in patients with diabetes. In dementia, cognition, and mood disorders, early excitement has not panned out in controlled trials. In AUD, we now have a positive RCT of semaglutide in patients with comorbid obesity, making it a reasonable consideration in this group. Data in BED are mixed.