The sample size is small enough that the correlation vs causation question still goes strong. You're arguing causation, the FDA argues correlation. Is it because they're taking the drugs, or because they're the type of person who would be taking those drugs and therefore have some kind of issue that needed help? It can be hard to know. Did they have these ideas or desires before taking the medication, or only after? If they are young, is it because of the meds, or because they're an angsty and hormonal teenager who has mental issues, or did the drugs exacerbate that?

It's easy to say "look at all these very public cases where this bad thing happened, and they all have this in common!", but that doesn't tip the needle for correlation vs causation at all imo.