I found the explanation for the $\lesssim$ in this question:
What does a tilde underneath an inequality mean?
My problem is of the following nature:
$g_t(y)$ is a stochastic function and $M_t$ some stochastic process, where $t\in[0,T]$ is the time parameter. Then there is the growth condition \begin{align*} |g_t(y)|\lesssim 1+|y|+|M_t|. \end{align*} Now, when squaring $g_t(y)$ I wonder, whether the growth condition expands to \begin{align*} |g_t(y)|^2\lesssim 1+|y|^2+|M_t|^2 \text{ or }\\ |g_t(y)|^2 \lesssim (1+|y|+|M_t|)^2. \end{align*}
The reason for my question is, that I found an inequation, where they used the first inequation. This triggered the thought, that maybe $g_t(y)$'s growth is determined in the different parameters independently and thus one can square them individually.