Cannabis Didn't Get Safer Than Alcohol — the Evidence Against It Just Got Sharper

Picture the version of “cutting back” a lot of people are living right now: fewer six-packs in the fridge, more 5mg THC seltzers on the counter, because the can says it’s “lighter than a beer” and every dispensary display has spent years selling the same pitch — weed is the grown-up, health-conscious swap. Dry January turns into dry-ish forever, minus the hangover, plus a gummy on weekends. It feels like harm reduction. Two studies published the same day, August 19, 2026, say the ground under that feeling just shifted, and not in the direction anyone marketing THC beverages wants you to hear.

The first, a JAMA Psychiatry analysis led by Beth Han and colleagues at the National Institute on Drug Abuse, surveyed 186,823 U.S. adults from 2021 to 2024 and found the substitution is real — cannabis use disorder is climbing while alcohol use disorder is falling or flat. The second, a meta-analysis in Biological Psychiatry led by Carly Stevens at UNSW Sydney and reported by ABC News Australia, pooled 149 studies of more than 71,000 people and found cannabis users develop psychosis earlier than non-users — and the gap gets wider, not narrower, with age.

Cannabis is not the safer swap for alcohol it is being sold as.

It’s a real substitution happening in real refrigerators, at exactly the moment the risk evidence on cannabis is getting more specific and more alarming, not less.

People are actually making the trade

Start with what’s changed in people’s habits. Among men in the NSDUH survey — the same federal data set behind the national opioid and alcohol trend numbers — past-year cannabis use disorder rose from 7.3% in 2021 to 9.3% in 2024. “Cannabis use disorder,” or CUD, is the clinical name for a pattern most people would just call losing control of your weed use: needing more to get the same effect, trying to cut back and failing, using it in ways that mess with work, relationships, or health. The moderate-to-severe version — four or more of those symptoms at once, the level closer to what people picture when they hear “dependence” — nearly doubled in men, from 3.1% to 4.1%. Women saw the same direction of travel: CUD from 4.5% to 5.6%, moderate-to-severe from 1.7% to 2.5%. Han’s team writes that the results “challenge the notion that cannabis has low addictive potential,” which, as the American Psychiatric Association’s own summary put it, “repudiate conventional assumptions that cannabis carries a low addiction risk” — a research-paper way of saying the industry talking point was wrong.

Alcohol moved the other way. Women’s alcohol use disorder — the DSM-5 term for problem drinking that meets clinical criteria, from mild to severe — fell from 9.7% to 8.2%. Men’s held roughly flat overall, dipping slightly from 13.3% to 12.4%. That’s the substitution story in one pair of lines on a chart, and it lines up with everything else going on culturally: legal cannabis in most states, sober-curious branding, THC drinks positioned on menus next to mocktails as the responsible choice. This isn’t a fringe behavior. It’s the mainstream version of moderation right now, and it shows up clearly in the science and medicine coverage tracking how substance-use patterns actually move once policy and marketing change around them.

The swap has a catch nobody put on the label

Here’s the part that should bother anyone treating cannabis as the automatically-safer choice. The one age group where alcohol use disorder actually rose was young men 18 to 20 — from 6.9% to 10.4% — even as the broader trend moved toward cannabis. And within that same 18-20 bracket, the JAMA Psychiatry data shows something sharper: among people who used cannabis, 47% to 50% met criteria for a use disorder, and more than half of those cases were moderate-to-severe. Among people who drank, only about 20% to 21% developed alcohol use disorder, and most of those cases were mild. More young people who try cannabis end up with a diagnosable problem than young people who drink — even though more of them drink in the first place. That is close to the opposite of “safer.”

More young people who try cannabis end up with a diagnosable problem than young people who drink — even though more of them drink in the first place.

None of this proves cannabis is worse than alcohol across the board — alcohol still kills people directly through overdose, organ failure, and withdrawal in ways cannabis essentially doesn’t, and this is a cross-sectional survey tracking national rates over four years, not a study following the same individuals over decades. It relies on self-report, which tends to undercount stigmatized use, and it can’t tell us whether someone’s state legalized recreational sales or how that shaped their answers. What it can tell us, cleanly, is that the story “cannabis carries low addiction risk” doesn’t hold up against the numbers people are actually reporting about their own lives.

The psychosis gap that grows with you

The UNSW findings add a second, separate warning that isn’t about addiction at all — it’s about timing. Across the pooled studies, cannabis users developed psychosis, on average, two and a half years earlier than non-users. Psychosis onset just means the age when someone has their first break from reality — hallucinations, delusions, the kind of episode that usually brings someone into psychiatric care for the first time. Under age 20, there was no significant gap at all, which cuts against the idea that this is purely a teenage-brain problem. But the gap opened fast after that: 1.6 years earlier in the 20-24 bracket, 4.43 years earlier at 25-29, and more than six years earlier for people who develop psychosis in their 30s. Matthew Large, the study’s senior investigator at UNSW, told the university’s newsroom plainly: “The risk doesn’t disappear once you’re in your 20s. What we’re seeing suggests a cumulative effect, where harm builds up over time, a bit like alcohol or nicotine.”

The honest caveat, which the researchers themselves state, is that this is a meta-analysis of observational studies — it cannot prove cannabis causes psychosis to arrive early, because the controlled experiment that would prove it would be unethical to run. Most of the underlying studies also focused on people already diagnosed with schizophrenia-spectrum illness, so the finding is strongest for that population and less certain for milder or one-off psychotic experiences. What the pattern does support, credibly, is that this isn’t a young-person’s risk you age out of. It compounds. That distinction matters specifically for cannabinoid users in their 30s, 40s, and 50s who assume the psychosis warnings are about teenagers and don’t apply to them.

The trade you’re actually making

Nobody is asking you to go back to the six-pack. But “safer than alcohol” was never the same claim as “safe,” and two studies landing on the same day just made the gap between those two claims impossible to ignore. If you swapped the drink for the drink-adjacent gummy because it felt like the healthier version of the same habit, the honest update isn’t to swap back — it’s to stop assuming either substance gets a pass just because the other one is worse.