According to this paper, Serre proved that there exists a pair $X, X'$ of smooth complex projective varieties, such that $X, X'$ are conjugate but not homeomorphic. Here, we say that $X, X'$ are conjugate if there exists a diagram
$\require{AMScd}$ \begin{CD} X @>{}>> X'\\ @VVV @VVV\\ \operatorname{Spec} \mathbb{C} @>{\sigma}>> \operatorname{Spec} \mathbb{C}, \end{CD} where $\sigma$ is complex conjugation.
I'm completely confused by this statement: complex conjugation induces a $\mathbb{Z}/2$ action on $\mathbb{CP}^n$ by homeomorphisms. By assumption, $X \subset \mathbb{CP}^n$, and the image of $X$ under this homeomorphism is precisely $X'$ (at least I think so...). Why does this not contradict Serre theorem?