What Instagram Leaves Out

Researchers collected semaglutide posts from Instagram, TikTok, X, Facebook and YouTube and compared what they said against what the STEP trials recorded.^[1] The most useful finding is not a false claim. It is an omission: the gastrointestinal effects that trouble most people who take this drug were not mentioned on Instagram at all.

What was missing

Misinformation is usually imagined as false statements — things a careful reader could catch. The pattern this analysis describes is different and harder to defend against: an account of a drug that is accurate as far as it goes, with the uncomfortable parts simply not present.

• Side effects. Nausea, vomiting, constipation and diarrhea are the commonest experiences on this drug and appear in every label. On Instagram, the analysis found them unmentioned.
• Cost. Barely covered, if at all — despite being among the most common reasons people stop.
• Supply shortages. Likewise, in a period when they were a defining feature of getting the drug.
• Indication. Semaglutide was presented almost exclusively as a weight-loss drug, with its diabetes indication largely absent.

Someone who learned about this drug entirely from Instagram would know what it does and not that it makes people sick.

The authors also report common misrepresentation of how the drug works, what it is for, its adverse effects and its interactions. Those are ordinary errors. The absences are the more consequential problem, because a reader cannot fact-check a topic that never came up.

Who is doing the explaining

On Instagram and TikTok, most of the content creators were women taking semaglutide themselves, without medical training. That is worth stating without condescension: firsthand accounts of living with a treatment are genuinely valuable, and people share them for good reasons.

What a personal account cannot do is describe the range. Someone posting has one experience of the drug; the label describes a distribution. If the people posting happen to be the ones who tolerated it — and those are the people still taking it, and therefore still posting — the aggregate picture skews without anyone saying anything untrue.

Across platforms, the analysis rated YouTube the most reputable source of medical information and Instagram the least.

The part that overlaps with what we track

The authors also identify violations of the Medicinal Products Act. One clarification matters for readers here: that is German pharmaceutical law, in a German-published analysis. It is not a US statute, and nothing in this paper represents a finding by a US regulator.

Finally, the authors judge that weight-loss result videos have a high probability of having been modified with AI — before-and-after imagery being both easy to generate and impossible for a viewer to verify.

How much weight to put on this

This is a content analysis, not a measurement. Phrases like very common and barely covered are the authors’ characterizations of the sample they gathered, and the published abstract does not give a sampling frame or counts. We are not going to convert them into percentages, because percentages would imply a precision the study does not claim.

Within that limit the direction is credible and matches what anyone who has looked at these platforms would expect. The specific, checkable observation — that a drug’s most common adverse effects were absent from one platform’s coverage of it — is the one worth carrying.

For what the trials actually record about side effects and how long they last, see how long GLP-1 side effects last. For the constipation nobody posts about, what actually works.