From where Yuval left off, it's easy to prove the case $p=2$. In that case
$$|(1-(\sin \alpha)^p)^{1/p} - (1-(\sin \beta)^p)^{1/p}| \leq (1 - (\cos (\alpha-\beta))^p)^{1/p} \; ,$$
reduces to
$$|\cos \alpha - \cos \beta | \leq \sin(\alpha-\beta) \; ,$$
assuming without loss of generality that $\alpha > \beta$. (Note, the case $\alpha=\beta$ is trivial even when $p>2$ anyway). With this assumption, we can also write
$$\cos \beta \leq \cos \alpha + \sin(\alpha-\beta) \; ,$$
or working out the $\sin$ with a formula for subtraction of angles and rearranging
$$\frac{\cos \beta}{1-\sin \beta} \leq \frac{\cos \alpha}{1-\sin \alpha} \; .$$
Thus, proving the case $p=2$ amounts to showing that $\frac{\cos x}{1-\sin x}$ is an increasing function of $x$ over $[0,\pi/2]$, which can be done by looking at its derivative.
I didn't find the case $p>2$ yet, but I still think exploiting the concavity properties of $(1-x^p)^{1/p}$ is a key to solving the problem.