GLP-1 drugs were supposed to make people eat less and lose weight. And they did. What was not understood was how far their behavioural effects might extend, because they have seemingly now gone much further.

I first started looking into this category of drugs when I was working on my paper on Nudgment, and the idea of how weak signals can emerge and in the aggregate, in combination, or over time, with observation, become much stronger strategic signals. One of the examples we used in that paper was the unexpected and unintended societal effects of GLP-1, (smaller portions, different snacking and grocery shopping habits) which were just emerging. They are now more than just emergent in some areas and new evidence is explaining in part why this may be happening.

When semaglutide and the other glucagon-like peptide-1 receptor agonists first became mass-market phenomena, their behavioural effects were understood primarily through appetite. People taking them became full faster. They thought about food less. They lost weight.

Then patients began reporting something stranger. They didn’t just want less food. Some wanted less alcohol. Some stopped smoking as much. Anecdotal reports described disappearing compulsions, reduced interest in shopping and other rewards, and, in some cases, changes in sexual desire. Researchers subsequently began finding signals in alcohol and opioid use disorders, while animal studies showed effects involving nicotine, stimulants and sexual behaviour.

What looked like appetite suppression started looking like something considerably more interesting: GLP-1 drugs may be weakening the urge to go after a reward. Scientists call this attenuation of reward-seeking, a recalibration of motivational salience. The emerging science does not suggest that GLP-1 drugs make previously pleasurable things less pleasurable. Instead, they may interfere with the process that turns a rewarding object into something we feel compelled to pursue.

That raises a much larger question than whether Ozempic can help people lose weight. What if GLP-1 drugs can alter wanting itself?

From Weight Loss Drug to Behavioural Drug

The discourse is extremely focussed on AI, in part because it follows the money, but it’s arguable that AI has yet to have a substantially positive impact on the human experience. Yet GLP-1 drugs may already have had a more direct positive effect on millions of lives, while their wider possibilities remain largely unexplored.

The expansion of GLP-1 effects has also happened remarkably quickly.

This is what some of the data is now suggesting. The alcohol findings are particularly important because anecdotes have now crossed into experimental medicine. A 2025 randomized phase 2 trial published in JAMA Psychiatry assigned 48 adults with alcohol use disorder to low-dose semaglutide or placebo. Semaglutide reduced the amount of alcohol consumed in a laboratory self-administration experiment, reduced drinks per drinking day and significantly reduced alcohol craving. It also produced greater reductions in heavy drinking over time. Among the small subgroup who smoked, cigarette consumption declined as well. (JAMA Network)

Then, in July 2026, researchers reported another randomized phase 2 trial, this time testing oral semaglutide in 50 treatment-seeking adults with moderate-to-severe alcohol use disorder. The existence of a second prospective randomized study is significant: GLP-1 treatment for addiction is moving beyond retrospective database mining and intriguing patient stories into a genuine clinical research program. (PubMed)

The evidence is not conclusive. A 2026 systematic review and meta-analysis identified only five randomized controlled trials involving 764 participants across substance-use outcomes, and its pooled results did not find statistically significant improvements in several alcohol and nicotine measures. The field is promising, but it is not settled. (PubMed) That combination, striking signals accompanied by inconsistent aggregate evidence, is exactly what should be expected from a research area this early.

The Interesting Part Is Dopamine

To understand why these drugs might affect behaviours that appear completely unrelated to eating, it helps to stop thinking about hunger.

Think about what wanting feels like. Dopamine is routinely described as the brain’s pleasure chemical, but that shorthand is misleading. Dopaminergic signalling is deeply involved in learning, anticipation, motivation, reinforcement and incentive salience: the process through which the brain assigns importance to something and says, in effect: Go get that.

GLP-1 receptors exist in brain regions involved in reward processing, including components of the mesolimbic system. Contemporary reviews increasingly connect GLP-1 signalling with the ventral tegmental area, nucleus accumbens and related circuitry governing reinforcement and motivated behaviour.

A major 2026 review of obesity and addiction research describes the overlap explicitly: obesity and substance-use disorders involve some of the same dysregulated mesocorticolimbic reward circuitry, while GLP-1 receptor agonists appear capable of acting on both metabolic systems and central reward pathways. Across animal studies, GLP-1 agonism has reduced intake or reinforcement involving alcohol, nicotine, opioids and stimulants. (ScienceDirect)

An even newer systematic evaluation, published August 21 in Current Psychiatry Reports, sharpens the hypothesis further. The emerging model is not that GLP-1 agonists simply turn dopamine down everywhere. Rather, GLP-1 signalling appears capable of attenuating aspects of cue-triggered, phasic dopamine signalling involved in reward seeking while leaving baseline dopaminergic function comparatively intact. (Springer Link)

The clearest experimental support for that distinction comes from a 2026 Nature study using mice engineered with the human version of the GLP‑1 receptor. Researchers found that the oral small-molecule agonists orforglipron and danuglipron engaged both the familiar circuits governing metabolic hunger and a separate population of neurons in the central amygdala. Activating that pathway selectively reduced consumption of palatable food by suppressing dopamine release in the nucleus accumbens. When researchers removed GLP‑1 receptors from those amygdala neurons, the drugs became less effective against reward-driven eating. It is unusually direct evidence that, at least in this animal model and with these particular compounds, GLP‑1 signalling can distinguish eating for energy from eating for reward.

That is a major difference. How we think about things we want, but don’t need, is in many ways, the pure essence of human motivation. We set goals and move toward those things. A promotion. Something material. A relationship. Something pleasurable, like winning a game, or eating something delicious when we are not hungry.

Think about passing a bakery. The smell of a cinnamon roll produces a cue. The cue predicts reward. The brain assigns the object motivational significance. Attention narrows. Anticipation grows. The person doesn’t know for certain that the cinnamon roll would taste good, but based on previous experience, they have a pretty good idea.

We want it. Now substitute: a cigarette. A glass of wine. Heroin. A slot machine. Sex. A shopping app. The underlying behaviours are obviously not identical. But each can recruit systems that transform cues into anticipation and anticipation into action. GLP-1 drugs may be interfering somewhere along that chain.

Pleasure and Wanting Are Not the Same Thing

This may also explain one of the strangest characteristics of GLP-1 anecdotes. People do not necessarily report that food suddenly tastes terrible. They report that they don’t think about it.

The distinction between liking and wanting has been fundamental to reward neuroscience for decades. A person can enjoy something when it appears without spending the preceding six hours thinking about obtaining it.

That suggests an unexpected interpretation of the GLP-1 phenomenon. Maybe the drugs don’t primarily eliminate pleasure. Maybe they reduce urgency. The anticipation still exists of the cake tasting good, but it exerts less pull, leaving more room for intention to overrule impulse. If that mechanism generalizes, the consequences extend well beyond obesity.

Addiction May Be the Obvious Next Frontier

Substance-use disorders are the most immediate application. The opioid evidence is already provocative. A 2024 JAMA Network Open study examined more than 33,000 people with both type 2 diabetes and opioid use disorder. Compared with several other diabetes medications, semaglutide was associated with substantially lower opioid-overdose risk over the following year, with hazard ratios ranging from 0.32 to 0.58 depending on the comparison drug. JAMA Network

That doesn’t prove semaglutide prevents opioid overdose. It was an observational study, and the authors explicitly warned about residual confounding and called for randomized trials. But combine it with animal experiments showing effects on opioid reinforcement and the human alcohol trials and an extraordinary possibility emerges.

GLP-1 agonists may eventually become treatments not for one particular addiction but for a component shared across several addictions: cue-induced wanting.That could make them fundamentally different from drugs developed specifically around alcohol, nicotine or opioids. They would target part of the machinery of compulsion rather than only the substance being compulsively consumed.

And Then Things Get Weird

Because drugs are not the only things humans compulsively pursue. Sexual behaviour is one example where the hypothesis is already being tested. Animal experiments have found that the GLP-1 agonist exendin-4 can reduce sexual interaction behaviours when administered into specific reward-related brain regions, including areas connected to the VTA and nucleus accumbens. Reviews published in 2025 and 2026 have consequently begun asking whether GLP-1 treatment might alter human libido. (PubMed Central (PMC))

But the human picture is complicated. Losing substantial weight can improve body image, hormonal function, vascular health, mobility and sexual confidence, all of which can increase sexual function even if reward-related sexual motivation is simultaneously being dampened. Recent reviews therefore find plausible pathways running in both directions. (PubMed) That complication points toward the much bigger research question.

What Counts as a Reward?

Many things, and it varies from person to person, circumstance to circumstance. As we acquire more, what we want changes, or the levels. Gambling certainly does. Shopping can. Social-media engagement, attention, status, romantic pursuit, all can. Approval. Revenge. Political outrage.

The neuroscience becomes considerably less straightforward as we move down that list. There is currently no evidence that semaglutide makes people stop loving their partners, abandon political ideologies or become less hateful. Those claims would be irresponsible. But there is now enough evidence about reward modulation to make the questions scientifically legitimate.

Could You Medicate Hate?

Hatred is not an addiction in the pharmacological sense. Neither are political extremism, romantic obsession or resentment. They are enormously complex states involving memory, identity, cognition, social reinforcement, threat perception, group membership, culture and moral judgement. Dopamine is involved in human motivation, but human motivation cannot be reduced to dopamine.

Still, some destructive behaviours acquire addiction-like characteristics. Online outrage offers a particularly interesting example. A person encounters a provocative stimulus, responds, receives social reinforcement, seeks another stimulus and repeats the behaviour. Recommendation systems can intensify the cycle by supplying increasingly potent cues.

A GLP-1 agonist is not a “hate drug.” But if these medications genuinely decrease cue-triggered incentive salience across multiple reward categories, researchers will eventually need to ask whether they affect compulsive social behaviours as well. That question has barely begun to be studied. The useful cut is not “hot vs cold hostility.” It is animus vs obligation. Both can be problematic, but they come from different motivations:

• Obligation: I will act against them because that is the job, the law, or the strategy.

• Hate: I want them reduced, and the wanting renews itself from cues.


Only the second is a candidate for a reward-circuit account. Only the second belongs in the same family as craving. The first can be morally worse in outcomes and still not be the same psychological object. We don’t want to speculate whether a drug could make people less obedient, less tribal, or less willing to carry out a policy. These are very fine lines of speculation with potentially dehumanizing outcomes many of which have been extensively explored in science fiction. But it is worth asking whether a drug that turns down cue-triggered wanting would reduce the appetite for another’s harm.

The Risk Is the Same as the Opportunity

There is an uncomfortable inverse to all this. A drug capable of reducing pathological wanting could potentially reduce healthy wanting too. Food addiction is undesirable. Wanting dinner is necessary. Alcohol dependence can be devastating. Wanting to celebrate with friends is not.

Sexual desire is also a biologically driven, necessary part of human experience. Like other addictions, compulsive sexual behaviour can destroy lives. The latest psychiatric review is important because it documents reports of depressed mood, anxiety, insomnia and emotional blunting emerging after GLP-1 treatment in some patients, sometimes improving after discontinuation. At the same time, the overall evidence does not establish a generalized GLP-1-induced anhedonia syndrome, and psychiatric outcomes across studies remain heterogeneous. (Springer Link)

That may eventually become the central therapeutic problem.

How do we suppress pathological salience without suppressing salience itself?

We don’t want medicines that make people indifferent to everything. We want medicines that allow people to decide what deserves their attention rather than having their attention captured for them.

Want and ambition are double-edged swords. They drive productivity, discovery, economic growth and personal change, but they can also produce obsession, exploitation, burnout and endless dissatisfaction. The converse is equally true. Less ambition might free people from the relentless pressure to achieve, but it could also remove the impulse that helps them act, create, solve problems or change circumstances that make them unhappy. For someone who still believes they should be accomplishing something but no longer feels the internal drive to pursue it, the gap between expectation and motivation could itself become psychologically distressing. The question is therefore not simply whether GLP-1 drugs reduce wanting, but which forms of wanting they reduce, by how much, and whether the person experiences that change as freedom or loss.

A Drug for Agency?

This is possibly the most interesting way to think about GLP-1 drugs. Obesity was the first obvious manifestation because food intake is measurable and weight falls dramatically.

But what patients repeatedly described was something more psychological. The food “noise” disappeared: thinking about food, being stimulated into food wanting by ads about snacks, thinking about decisions about food, contemplating what to eat next even without hunger. That phrase may eventually turn out to have been telling us what these drugs actually do.

Noise is not pleasure. Noise is intrusion. It is the thought that keeps returning, the drink you promised yourself you wouldn’t have, the cigarette you quit three days ago. The food in the refrigerator at midnight. The purchase you don’t need. The notification you cannot resist opening. The promotion you might not get.

Humans spend an extraordinary amount of their lives negotiating with impulses they did not consciously choose to generate. For centuries, we have treated that struggle primarily as a problem of character, discipline or willpower.

GLP-1 research is beginning to suggest that at least some of what we call willpower may instead be the downstream experience of biological systems assigning different strengths to competing rewards. That does not mean semaglutide is going to cure addiction, compulsive shopping, gambling, obsessive romantic attachment or hatred.

The evidence gets progressively thinner as we move away from food and substances, and for several of those behaviours it does not yet exist at all. But the trajectory is extraordinary. A hormone pathway studied for glucose regulation led to diabetes drugs. Those drugs became obesity drugs. The obesity drugs unexpectedly became probes into appetite, reinforcement and addiction.

Now they are giving neuroscience an unusually powerful experiment in what happens when human wanting itself is turned down. There is also a less futuristic consequence already worth considering: what happens to an economy built around appetite when millions of people become less appetitive? The first-order commercial effects are relatively easy to imagine and, in some cases, are already being experienced by food companies and restaurants: smaller portions, fewer impulse purchases, different menus and declining demand for highly rewarding foods and alcohol. But if reduced incentive salience eventually proves to extend meaningfully beyond food and substances, the second-order effects could be much larger. People who want less may buy less, browse less, gamble less and respond less readily to advertising designed to manufacture craving. More consequentially, if future research were to establish reductions in social motivation, sexual pursuit or romantic desire, which it emphatically has not established yet, the effects would reach dating, nightlife, hospitality, travel, entertainment and ultimately relationship formation itself. At population scale, even a modest reduction in the desire to go out, meet people, pursue partners or seek novelty could matter in societies already struggling with loneliness, declining marriage and partnership rates, and falling fertility.

The individual experience of being liberated from an unwanted craving can be profoundly positive; an entire society becoming slightly less motivated to eat, drink, shop, socialize, explore, court and connect is a very different proposition. GLP-1 drugs therefore raise an unusual economic question: what happens to a consumption economy, and perhaps a social one, when wanting becomes less abundant? The implications are philosophical sociological and existential if we consume less the strain on the environment lessens, but our economy changes as well. These are enormous adjustments to consider and at times the contemplation feels like something out of a science fiction movie, like so much else we’re living through right now.

There is another implication hiding inside this research that is considerably more unsettling. If scientists are identifying biological mechanisms through which the salience of rewards can be turned down, the obvious inverse question is whether those same systems can eventually be manipulated to turn wanting up. That possibility has received far less attention. A sufficiently precise understanding of the circuitry that determines whether a cue feels irrelevant or irresistible could theoretically be used not merely to suppress craving but to amplify it, to increase motivation, attachment, consumption or compulsive pursuit. That is still speculative; there is no GLP-1 “desire amplifier” waiting in a laboratory. But the direction of the research is relevant.

Exploring impact on negative motivations might generate even more useful, fruitful results. Hatred is not a substance-use disorder. But some of the behaviors that keep it running, revenge, schadenfreude, and the online outrage loop, do recruit the same reward circuitry that assigns incentive salience to a cue. The grievance hurts; anticipated retaliation or shared condemnation is tagged as worth pursuing. That is a reward process. It is also only one layer. Identity, threat, and narrative can maintain hostility after the craving-like burst is gone. That is why a drug that turns down cue-triggered wanting might, in principle, blunt compulsive payback seeking without touching the belief that the other side deserves it, and why that remains a research question, not a finding.

Modern advertising and recommendation systems already devote enormous computational resources to discovering which cues maximize human engagement. Pharmacology capable of altering the strength of the response to those cues would introduce an entirely different category of behavioural influence. The benign version might help restore motivation in people suffering from pathological anhedonia. The malign version is considerably easier to imagine: an economy already extraordinarily good at manufacturing desire acquiring tools capable of changing the biological gain on desire itself. The discovery that wanting can potentially be dialled down should therefore provoke a second question almost immediately: who eventually gets access to the dial?