Ehhhhhhh...
And google swings... and it misses...
There are two somewhat competing studies involved in what you linked.
The original meta-analysis is stating that "[a]ll antidepressants were more efficacious than placebo in adults with major depressive disorder." And if you dig into that study, the level of effectiveness varies, but is consistently above a placebo.
The other, states that the original meta-analysis missed some points within their study; specifically that under the standardized criterion, there is an unknown level of bias in the results (73% showing unknown bias, and 26% showing high risk of bias - essentially confirming that most studies couldn't be done in a full blind or double-blind manner); and that while the sample size was high, the effect size was less known due to drop-out rates of the studies.
This isn't necessarily contrary to the initial meta-analysis, but brings to light that they may need to be taken with a level of caution; it doesn't make the claim that anti-depressants are no better than placebos, but that when doing the meta-analysis Cipriani et. al. needed to make more apparent exigencies with their results.
All this being said, the puff piece draws a more black and white tone with the meta-analysis and the meta-meta-analysis results; which tends to be the problem with these types of pieces. Sure, do your own research, but you'll need to dig into what's actually behind the journalistic fluff to get at the real meat and potatoes of the subject, if you don't you may come away with the idea that anti-depressants have 0 efficacy instead of "some efficacy, with the caveat that studies are somewhat mixed on individual results due to the nature of how the studies are conducted."