Nothing guarantees that the derivative of the curve defined by @Yiorgos is integrable and this makes his solution questionable. I give an alternative solution below.
For each $x\in \Bbb R^m$, define the curve
\begin{align*}
\begin{array}{cccc}
F_x:&\Bbb R&\to&\Bbb R^n\\
&t&\mapsto&|x|f(tx)
\end{array}.
\end{align*}
For each $x$ fixed, $F_x$ is differentiable, since it is the composition $t\mapsto tx\mapsto f(tx)\mapsto|x|f(tx)$. Furtheremore, $F_x'(t)=|x|f'(tx)\cdot x$, by the chain rule.
Since $\displaystyle \lim_{|x|\to \infty}f'(x)\cdot x=0$, there exists $M>0$ such that $|x|>M\implies |f'(x)\cdot x|<1$.
Of course, $g(x)=f(2x)-f(x)$ is bounded on $B[0;M]$ (because it is continuous and the closed ball is compact).
It remains to show that $g$ is bounded on $\Bbb R^m-B[0;M]$. Let $x\in \Bbb R^m-B[0;M]$, i.e., $|x|>M$. For $t\in [1,2]$, we have $|tx|=|t||x|\geq |x|>M$ and, therefore, $|f'(tx)\cdot tx|<1$, for all $t\in [1,2]$. Then we have $|F_x'(t)|=||x|f'(tx)\cdot x|=\frac{1}{t}|x||f'(tx)\cdot tx|\leq |x|$, for all $t\in [1,2]$. By the Mean Value Inequality applied to $F_x$ over $[1,2]$, we get
\begin{align*}
|x||g(x)|&=|x||f(2x)-f(x)|=|F_x(2)-F_x(1)|\leq |x||2-1|=|x|,\end{align*}
and this implies that $|g(x)|\leq 1$, if $|x|>M$.
This proves that $g$ is bounded on $\Bbb R^m$.