Yea, there is quite a lot of survivorship bias in this comparison. "People who succeeded at controlling hunger and improving diet" vs "People who stopped taking their medicine". It is very unsurprising.
I'm not the best at reading medical studies, but it seems to me like they are taking averages over all subjects in the randomized controlled trials, not just "successful" subjects.
A "fair" comparison would be to compare weight gain of people who lost weight naturally, with Ozempic users who _would have_ lost weight naturally, but it's not really possible to know which Ozempic users fall into that category.
The study compared "everyone who did this intervention" vs " everyone who did this other intervention". That seems perfectly fair to me, it's just answering a different question to your proposed comparison.