Nothing has established that they do. What exists is a count of reports. Across 42,941 European adverse event reports from 2021 to 2025, suicide or self-injury appeared 37 times for liraglutide. For semaglutide it was 141, and for tirzepatide 47.
Relative to tirzepatide, semaglutide reported these events 2.69 times as often, 95% CI 1.91 to 3.83. Liraglutide reported them 2.54 times as often. European and US regulators have not confirmed a causal link, and the authors say this analysis does not establish one.
The cohort evidence runs the other way. A Swedish national study followed 95,490 people who already had depression or anxiety[2]. Semaglutide was associated with a lower risk of worsening mental illness, aHR 0.58. Liraglutide came to 0.82. Exenatide and dulaglutide showed nothing, and two of the four drugs did not move the outcome at all.
These numbers describe reports, not people. That distinction is the whole article. It is set out at length in the piece on the FDA’s database. That one runs a different comparison on a different system.
What was counted
Every European case safety report from 2021 through 2025 naming one of the three drugs as suspect. The indication had to be diabetes or weight management. [1] That came to 42,941 reports. Gastrointestinal problems dominated, as they always do for this class.
Suicide and self-injury events were a small share: 37, 141 and 47 cases across the three drugs. 225 in total.
The comparison
The reference drug is tirzepatide. Not placebo, not every other medicine in the database — tirzepatide.
Against it, semaglutide carried these events 2.69 times as often among its reports, 95% CI 1.91 to 3.83. Liraglutide, 2.54 times, 95% CI 1.60 to 4.01.
What else runs the same direction
Liraglutide has been sold since 2010, semaglutide since 2018, tirzepatide since 2022. Older drugs accumulate reports.
And the suicidality question was a headline about semaglutide in particular. People report what they have been told to watch for. Pharmacovigilance calls that notoriety bias. It is the standing weakness of every spontaneous system.
What regulators found
European and US agencies have both examined this and neither has confirmed a causal relationship. The authors note that their results do not contradict that.
A large matched cohort pointed the other way, associating semaglutide with lower neuropsychiatric event rates. Though that study published no effect sizes at all. Neither design settles it.
What a buyer does with it
Not choose a molecule on this. The two drugs differ on weight and on price in measured ways. Here they do not differ in any way a rate could confirm.
If you are taking one of these and having thoughts of harming yourself, speak to a doctor now. This is not a number to weigh. Continued monitoring is what the authors ask for, and it is the right ask.