A Can of THC Seltzer Looks Like a Can of Seltzer. That Is the Whole Problem.

A July 21, 2026 investigation by STAT documented hemp-derived THC beverages displayed among soda, tea, coconut water, and chocolate milk across ten stores in South Brooklyn, with little shelf signage identifying them as intoxicating and most stores not asking for identification at checkout. Most cans were labeled at 5 to 10 milligrams of THC. Some carried 50, 60, or 100 milligrams.

New York’s own cannabinoid hemp rules limit these beverages to 1 milligram of total THC per serving, with one serving per package. A 10 milligram can is ten times that limit and a 100 milligram can is a hundred times it. The clinical point is not that every hemp beverage is dangerous. It is that a caregiver has no reliable way to tell a 1 milligram product from a 100 milligram product by looking at the cooler.

Liquids are consumed faster and in larger volumes than gummies, and a can carries no obvious dose boundary. A child who eats one gummy has eaten one gummy. A child who drinks a can has consumed whatever the can contained, and the label is the only thing that says what that was.

The recognition problem also reaches adults. Reviewing an exposure history with a patient who is sedated, confused, or vomiting, clinicians ask about gummies and about smoking. A can in a shared refrigerator is easy to miss on the first pass, and on the second pass the patient may sincerely not know it contained anything.

The most current national picture comes from an analysis published in the Journal of Addiction Medicine in June 2026 by Raymond Bertino, Ryan Feldman, and Jerrold Leikin, who reviewed National Poison Data System reports from 2000 through 2024. Exposures among children under six were stable through 2009, then began climbing. Between 2009 and 2024 they rose from 132 to 8,430, an increase of 6,386 percent. Among children aged six to twelve the count went from 21 to 2,894.

A separate analysis in Pediatrics by Marit Tweet, Antonia Nemanich, and Michael Wahl looked specifically at edible cannabis exposures in children under six from 2017 through 2021. It counted 7,043 exposures across those five years, rising from 207 in 2017 to 3,054 in 2021. Of those, 97.7 percent occurred in a residential setting. Among cases followed to a known outcome, 70 percent involved central nervous system depression, and 22.7 percent of all reported cases were admitted to a hospital.

Those are the real numbers and they deserve to be read carefully in both directions. Eight thousand calls a year across the entire United States is roughly twenty-three a day, and a poison center call is a call, not a confirmed poisoning. At the same time, an admission rate near one in four among reported edible exposures is not trivial, and central nervous system depression in a toddler is a serious presentation.

Neither analysis isolates beverages. The National Poison Data System records a cannabis exposure and, in the edible analyses, an edible product category. It does not carry a field that separates a seltzer from a gummy, and there is no national product-specific surveillance for hemp-derived THC drinks.

This matters because it cuts against overstating the case. Nobody can currently say what share of the rise in pediatric cannabis exposures is attributable to beverages, and anyone who tells you otherwise is guessing. What can be said is narrower and still useful: beverages are a format with no natural dose boundary, packaged in containers that resemble ordinary drinks, and increasingly stocked alongside them.

The STAT reporting is a journalistic spot check of ten stores in one part of one city. It cannot establish a violation rate for New York, let alone the country. It does establish, with specific observations, that the recognition problem exists in ordinary retail rather than only in specialty shops.

New York’s cannabinoid hemp regulations set the limits for products sold outside the licensed adult-use cannabis system. Orally consumed cannabinoid hemp products, a category that includes food, supplements, and beverages, are limited to no more than 1 milligram of total THC per serving and 10 milligrams per package, and may not exceed 0.3 percent delta-9 THC concentration.

Beverages carry an additional rule. A cannabinoid hemp beverage may contain no more than a single serving per package. The state’s guidance for licensees gives the arithmetic directly: one single serving can is one package, a six pack of single serving cans is six packages, and each can may contain no more than the maximum allowed in a single serving. In practical terms, a lawful New York cannabinoid hemp beverage tops out at 1 milligram of THC per can.

The regulations also prohibit cannabinoid hemp products containing synthetic cannabinoids, artificially derived cannabinoids, or cannabinoids created through isomerization, naming delta-8 and delta-10 THC specifically. A 10 milligram delta-9 seltzer or a delta-8 drink sold from an ordinary bodega cooler is not operating inside that framework.

Section 781 of Public Law 119-37, the appropriations act signed November 12, 2025, rewrote the federal definition of hemp in the Agricultural Marketing Act of 1946. It replaces the delta-9 threshold with a total tetrahydrocannabinols standard inclusive of tetrahydrocannabinolic acid at 0.3 percent, excludes products containing cannabinoids a Cannabis sativa L. plant cannot naturally produce, excludes cannabinoids the plant could produce but that were synthesized outside it, and caps final hemp-derived cannabinoid products at 0.4 milligrams of combined total THC and similar-acting cannabinoids per container.

On September 2, 2026 the president signed H.R. 6500, the Continuing Appropriations and Extensions Act, 2027. Its Section 2019 provides that until December 11, 2026 the Section 781 amendments apply only to the exclusions for cannabinoids not capable of being naturally produced by the plant. Everything else, including the 0.4 milligram per container cap, begins December 11, 2026.

Set that against a cooler. A can labeled 5 milligrams is more than twelve times the coming federal per-container limit. A can labeled 10 milligrams is twenty-five times it. A 100 milligram can is two hundred and fifty times it. Whatever happens to enforcement and to the inevitable litigation, the products that created this recognition problem are not designed to survive that number, and families should expect the shelf to look different in the first months of 2027. For the fuller timeline, including which exclusions were not delayed, see our explainer on the federal hemp deadlines.

For families, the instructions are short and they do not depend on how any of this litigation resolves. Keep intoxicating beverages in their original container so the label stays with the product. Do not store them in the same place as ordinary drinks, which means a different shelf at minimum and a locked location where children or visiting children are present. Read the total milligrams in the container rather than a per-serving figure, because with a beverage those are frequently the same number and occasionally are not. Do not drive after use.

If a child may have consumed a THC product, call Poison Control at 1-800-222-1222 in the United States. Call emergency services for trouble breathing, a seizure, repeated vomiting, severe confusion, or difficulty waking the child.

For clinicians, the practical change is to the exposure history. When a young child presents with unexplained lethargy, ataxia, vomiting, or behavioral change, ask about beverages by name alongside gummies and baked goods, and ask what is in the refrigerator rather than what is in the medicine cabinet. In adolescents and adults, ask the same question about drinks bought at a convenience store or gas station, where the person may not have registered the product as intoxicating at all.

The exposure trend is well established. Two independent analyses of the same national reporting system, covering different windows and different product categories, both show a steep and sustained rise in pediatric cannabis exposures that begins around 2010 and continues through the most recent year of available data. For the question of whether this is happening, the evidence is solid.

The regulatory facts are stronger still, because they come from published regulation and statute with citations and effective dates rather than from inference. What is weak is the link between the two. No dataset connects a specific beverage on a specific shelf to a specific emergency visit, and the honest position is that the recognition risk is plausible and documented in kind, not quantified.

Poison center data are a reporting stream, not an incidence measure. A call is generated when a caregiver notices something and decides to call, which means the numbers move with awareness, with media coverage, and with how easy it is to reach a poison center, not only with true exposures. Rising legalization also raises the chance that a caregiver will identify cannabis as the cause rather than reporting an unknown ingestion. Percentage increases from a small 2009 baseline look dramatic for arithmetic reasons and should be read alongside the absolute counts.

Ten store visits in one Brooklyn neighborhood cannot support a claim about national retail practice, and the reporting did not independently test the cans against their labels. A reader should treat those observations as a demonstration that the problem occurs, not as a measurement of how often.

None of this shows that THC beverages caused the rise in pediatric cannabis exposures, that they are more dangerous per milligram than any other oral format, or that a labeled 10 milligram drink produces the same effect as a labeled 10 milligram gummy. Formulation, emulsification, food in the stomach, and individual pharmacokinetics all differ, and the beverage-specific pharmacokinetic literature is thin.

It also does not show that the coming federal per-container limit will be enforced on any particular schedule, survive litigation intact, or change what a given state permits inside its own licensed cannabis system. Section 781 rewrites the federal definition of hemp. It does not govern state-licensed cannabis products.

The hemp-derived intoxicant market grew in the space between a federal definition written for an agricultural commodity and a consumer market that developed around it. Beverages were the format that carried that market into ordinary retail, because a can fits on a shelf that already exists, in a store that already sells drinks, to a customer who already buys them.

That is what makes this a labeling story rather than a pharmacology story. The active ingredient is not novel. What is novel is a psychoactive dose arriving in a package indistinguishable from a package with no psychoactive dose in it, sold in a place where nobody expects to encounter one. Federal law is about to address the dose. Nothing currently on the calendar addresses the package.

I do not think THC beverages are an emergency, and I do not think parents should be frightened by a percentage increase calculated off a 2009 baseline of 132. What I think is that we have built a product category where the safety of a household depends entirely on whether an adult reads six point type on a can.

In clinic the version of this I see most often is not a poisoned toddler. It is an adult who drank two of something at a cookout, had no idea what was in it, and arrived convinced something was seriously wrong with them. That is a preventable experience, and the prevention is not complicated. Keep these in their original container, keep them off the shelf where the ordinary drinks live, and lock them up if children come through your house. If you cannot tell what is in a can by reading the can, do not keep it in the refrigerator.

Hemp-derived THC beverages are sold in ordinary retail at doses that exceed what New York’s cannabinoid hemp rules permit, and a federal cap of 0.4 milligrams per container takes effect December 11, 2026. Until the shelf changes, treat every THC drink in a household as a medication: original container, separate storage, locked when children are present, and read the total milligrams before anyone opens it.

The finding to carry forward is that pediatric cannabis exposure reports have risen steeply and consistently for fifteen years, and that beverages are a format built for misrecognition. The claim not to carry forward is that beverages are driving that rise, because no surveillance system currently in place could tell you that.